Specialization Program by the numbers

With its vast and interconnected data, CMS Glance requires a smarter way for users to discover and access information. The AI-Driven Smart Search project aims to provide a unified, intelligent search system that spans all Glance modules—including analyses, members, institutes, notes, and tools. Using semantic search and AI embeddings, the system understands the intent behind queries, delivering contextually relevant results…

Every CMS physics analysis is steered from first draft to publication through a 17-stage editorial workflow in the Glance platform, where reviews stall and queues form. This project builds a system that measures where analyses pile up and predicts how long each one will wait, early enough to act on. The work has three parts: a forensic reconstruction of Oracle…

A CBCT scan is a 3D X-ray of the jaw, and today a surgeon writes its report by hand, one scan at a time. We're automating that, with one deliberate constraint: the AI that writes the sentences never sees the scan. A first stage finds and labels the anatomy (each tooth, the nerve canals, the sinuses, the jawbone), a second…

This project develops a spatial agent-based model of early influenza infection in human lung tissue, using Vivarium to simulate viral spread, immune-cell interactions, and diffusible signaling fields. The model is designed to connect computational simulations with CODEX, H&E, and viral titer data.

A tool that checks how well a Saudi company follows the national cybersecurity standards, using nothing but a scan of what the company already shows to the internet. No passwords, no software to install, no security team needed. It goes through every control in the ECC-2:2024 and NCNICC-1:2025 standards, sorts out which ones a scan can actually prove, and produces…

AnswerMap is a training-free, black-box method for spatial interpretability in vision language models. Given an image and any free-text query, it produces a heatmap over that image showing where the model itself locates the evidence, from which two read-outs are taken: a single point (where) and a scalar (how strongly the model commits to that region). The map is elicited…

Classify spoken Arabic into one of 11 country-level dialects under cross-domain conditions, where the test set differs in domain and source from training and development data. For domain generalization, a secondary-domain development set is provided alongside the primary development data.

Regularization in its various forms has been extensively used in neural networks, particularly in unsupervised learning models such as autoencoders. Recently, Mounayer et al. introduced Rank Reduction Autoencoders (RRAEs), demonstrating both theoretically and empirically that the Singular Value Decomposition (SVD) can serve as a powerful regularizer for autoencoders without requiring the optimization of additional loss functions. While this archiitecture shows…

Large Language Models (LLMs) are good at judging relevance but have a quadratic attention cost, and a limited context window. CorpusProbe factorizes the document to reduce the context window while also producing a subquadratic computational cost. We factorize a document into K chunks and pass through 2K yes/no single-pass probes, with the first K covering local context and the second…

Human pluripotent stem cell-derived cardiomyocytes (hPSC-CMs) are valuable for disease modeling and regenerative medicine but remain limited by functional immaturity. We developed a deep learning framework using a Deep Count Autoencoder with a Zero-Inflated Negative Binomial loss to extract robust transcriptomic maturation signatures from scRNA-seq data without log normalization. By minimizing batch effects between in vitro iPSC-CMs and adult GTEx…

KAUST PDEC (Planning, Design, Engineering, and Construction) is the university's in-house owner's representative consultancy, which plans, designs, and delivers KAUST's construction works. Our project is an AI-powered screening tool that reviews PDEC's scope-of-work documents (project specifications & BOQs) against the KAUST Standards, flagging each finding as inadequate, adequate, or exceeding compliance, with shortfalls and suggested solutions. It adds a compliance…

This project investigates modernising the CMS experiment's documentation system at CERN by migrating approximately 44,677 legacy TWiki pages to GitHub Flavored Markdown, a modern format compatible with CERN's Git-based infrastructure. After evaluating Pandoc and identifying its limitations, we develop a custom converter achieving 98.6% success on the full corpus. The project also implements an AI-powered semantic search pipeline with Retrieval-Augmented…

There are two established ways to write down a small, readable model of how an animal or a person makes repeated decisions. A Finite State Controller (FSC) stores what matters as a handful of labelled memory states and a lookup table. A tiny recurrent neural network (RNN) stores the same thing as a short list of numbers updated by trained…

Our project develops a deepfake and AI-generated image detection system that uses a vision–language model to provide both a verdict and a human-readable explanation. We fine-tuned the FakeReasoning model using memory-efficient QLoRA while preserving its pretrained visual features. The resulting system improves detection performance on our target dataset, maintains strong generalization across unseen image generators, and is deployed through a…

This project develops a high-level neural-network-aware communication framework for distributed inference across physically separated and resource-constrained IoT devices, enabling each device to act as an individual computing node within a neural network. The framework provides mechanisms for device addressing, message formatting, input synchronization, transmission scheduling, transport abstraction, and lightweight fault tolerance, allowing multiple IoT devices to coordinate a distributed forward…

DiacriticS is a research project focused on supervised fine-tuning (SFT) of open-weights language models to achieve robust Arabic Text Diacritisation (ATD) under strict contamination control. The project utilises a training corpus derived from Sadeed Tashkeela (~53 million words) with minimal overlap (0.4%) against evaluation data, while benchmark evaluations are conducted on SadeedDiac-25, a contamination-controlled dataset of 1,200 curated, human-reviewed paragraphs…

Vector search has become a core operation in modern retrieval systems, but many scientific and interactive workloads do not fit the top-k model that most systems are built around. Given a query object, a similarity radius, and optional filters over structured attributes, these workloads require every matching record to be returned, a threshold query, whose answer size varies from one…

In complex scientific detectors, reconstruction software tries to infer what happened during a particle collision starting from low-level sensor signals. At the same time, simulation provides a detailed ground-truth description of the event, including the particles that were generated and how they propagated through the detector. A powerful way to represent both reconstructed and simulated information is through graphs, where…

This project aims to evaluate potential energy generation scenarios for Saudi Arabia in preparation for hosting the FIFA World Cup 2034. The objective is to investigate how different energy mixes, including natural gas and renewable energy resources, could satisfy the expected increase in electricity demand while maintaining reliable and cost-effective system operation. The project will utilize the PLEXOS energy market…

An experimental study of three non-contact water level sensors Ultrasonic, Radar, and LiDAR to evaluate their performance, limitations, and overall reliability for monitoring rivers in remote areas without cellular coverage. As a complete solution, the system is designed to be low-cost and battery-efficient system, with a long operational lifetime that makes it well suited for remote river deployments. Sensor data…

Bone fracture classification from X-ray images is an important application of deep learning for supporting clinical diagnosis. However, high-capacity models are often computationallyexpensiveanddifficulttointerpret,limitingtheirdeploymentinresource- constrained healthcare environments. This study investigates an Explainable Knowledge Distillation(EKD)frameworkthatcombinesprediction-basedknowledgedistillationwith class-discriminative attention transfer to develop a lightweight and interpretable fracture classifier. Using the FracAtlas dataset, EfficientNet-B6 was selected and optimized as the teacher model, while MobileNetV2 was…

FAHIM is an intelligent, voice-driven Android assistant designed to help blind and visually impaired Arabic-speaking users complete multi-step tasks in mobile applications through natural Arabic voice interaction. AI-Driven Planning: Users express their goals through natural Arabic voice commands. The Android application captures and normalizes the current interface using Android Accessibility Services, then sends the user's request and the observed UI…

The province of Ontario in Canada has one of the greatest densities of lakes in the world. Sustaining and managing these important populations is vital for maintaining the ecosystem and allowing species persistence despite harvesting. A key measure that fisheries managers require is species abundance. This allows them to understand how abundances change spatially and temporally in response to various…

Fusion-JEPA is an advanced text-to-speech architecture that pairs a Joint-Embedding Predictive Architecture (JEPA) with a Multimodal Diffusion Transformer (MM-DiT) and continuous Flow Matching to synthesize high-fidelity speech in both Arabic and English.

Late-interaction models such as ColPali achieve strong visual document retrieval performance but require storing many token embeddings per page. We propose compact generative codes that capture each page's low-dimensional token manifold. The project has two central components: efficient retrieval and on-demand vector generation. The codes can be scored directly using standard MaxSim for first-stage retrieval, then used to generate richer…

We benchmark ten autoencoder regularization techniques — Vanilla, L1, L2, Contractive, Sparse, VAE, IRMAE, LoRAE, RRAE, and Denoising — on four synthetic PDE datasets of increasing difficulty: a translating Gaussian bump, thin-plate deflection, Gray–Scott reaction–diffusion, and a multi-channel shallow-water wave. All techniques share one convolutional backbone, one reconstruction loss, and one training budget, so differences come only from the penalty.…

The Arabian leopard (Panthera pardus nimr) is critically endangered, with only a small remnant population surviving across a vast range. Conservation depends on knowing where those animals still are, but the terrain is remote and off-grid, putting real-time detection beyond the reach of cloud infrastructure. We present a split inference architecture that distributes YOLO26L computation across two Raspberry Pi devices:…

Dark matter cannot be observed directly, so its distribution is studied through its gravitational effects on visible matter. This project uses synthetic dwarf galaxies and combines graph neural networks with Neural Posterior Estimation to infer the posterior distribution of dark matter density-profile parameters from five-dimensional stellar observations: two sky-position coordinates, line-of-sight velocity, and two proper-motion components. The graph neural network…

The Phase-2 upgrade of the CMS detector at the CERN Large Hadron Collider will require a new generation of reconstruction algorithms capable of handling the much higher event complexity and particle densities expected in future running conditions. In the high-granularity endcap calorimeter, the reconstruction of particle showers and the association between tracks and calorimeter tracksters are especially challenging, since the…

The Gaia mission is transforming astronomy by constructing a 3D map of nearly two billion stars in the Milky Way and beyond. Despite this unprecedented dataset, the scientific value of stellar spectra remains underutilized due to the lack of large-scale interactive systems for spectral exploration. We present LacteaChat, a project that extends the open-source Gaia spectral visualization system with a…

Estimating realistic indoor lighting from a single photograph remains challenging, because lighting is entangled with material and geometry. Previous methods still do not convey how the consistency and reflection of light is in real life or is computationally expensive. Luma aims to bridge the gap in existing light estimation methods by introducing a physics-grounded model trained with a physically-based differentiable…

Our project investigates how jailbreak attacks affect diffusion language models in healthcare. We study whether unsafe template injections leave detectable internal signatures and whether these signals can be used to steer the model back toward safer responses. Our findings show that harmful injections can be detected reliably, but correcting the resulting behavior remains more challenging.

The goal is to build a trustworthy, explainable, multi-agent research system that evaluates recorded interview answers and gives participants useful feedback instead of only an unexplained score.

This project develops a deep-learning pipeline that estimates the articulated pose of surgical instruments directly from Optical Coherence Tomography (OCT) volumes, enabling fast, image-based feedback for robotic microsurgery. Each OCT acquisition is represented as a sparse 3D point cloud and processed by VoxelNeXt, a fully sparse convolutional network that localizes eight 3D keypoints of a four-link tool model (base, shaft,…

This research paper investigates a generative autoencoder that enables both image reconstruction and generation without relying on the KL divergence used in standard Variational Autoencoders (VAEs). The proposed approach combines SVD-based latent regularization with multiplicative Gaussian noise to structure the latent space while maintaining effective reconstruction and generation capabilities.

A novel data-filtering mechanism designed to automatically detect and redact personal information from audio inputs before cloud processing occurs, which demonstrates the feasibility of masking private information without impeding the performance of AI Home Assistants. By ensuring that private conversations remain localized, this solution effectively mitigates the risk of external data leaks. Ultimately, this framework provides a practical safeguard tailored…

Efficient planning is necessary to ensure sustainable energy utilization with the rapid expansion of data centers and the increasing need for a reliable digital infrastructure. This project is intended to optimize the siting of data centers in Saudi Arabia by evaluating suitable locations and evaluating potential energy mix scenarios using the PLEXOS modeling platform. In the first phase of the…

Truffles, specialty coffee, and premium spices are foods of high value with a complex identity that is easy to imitate. This combination makes them frequent targets of fraud: a lower value item is sold as a premium one, or an aged sample is sold as fresh. Verifying such a claim today requires either expert judgment or a laboratory technique such…

Digital human motion contains many body positions for every second, making it expensive for AI models to store and process. Normal compression can lose important details, causing the reconstructed character to drift, slide its feet, or move incorrectly. PIAE compresses every eight frames while preserving the character's path, turning, height, and foot contact. When tested on 4,210 HumanML3D motions, PIAE…

Reconstruction software for the CMS experiment must run on GPUs to keep pace with the data rates of the High Luminosity LHC. Efficient GPU execution requires data laid out as a Structure of Arrays (SoA) rather than an Array of Structs. This project converts the Trackster a core data structure of the TICL reconstruction framework to an SoA layout using…

Modern astronomy produces far more images than can realistically be labeled by humans. Even large citizen-science efforts such as Galaxy Zoo have required years of volunteer effort to label only ~million galaxies, while reliable expert labels for rare astronomical objects are far scarcer. At the same time, vision-language models (VLMs) can perform image-classification tasks directly from text prompts, without task-specific…

Quantum Graph Learning for Jet Identification: This project investigates whether quantum computing can enhance how the LHC identifies jets — collimated sprays of particles from proton collisions — by extending a classical Graph Neural Network (ParticleNet) with a trained, compact latent representation of each jet, then encoding that representation onto Pasqal's neutral-atom quantum hardware via a custom-built compiler, in order…

ReadEnough is a training-free early-stopping layer for sequential long-context reading. The input is divided into fixed-order chunks, and a frozen language model maintains compact running notes that are updated after each chunk. For multiple-choice tasks, a short probe extracts the model's exact probability distribution over the answer options. ReadEnough accumulates evidence of distributional stability while requiring sufficient answer confidence, stopping…

Proximal Policy Optimization (PPO) relies on advantage estimates to guide policy updates, while Generalized Advantage Estimation (GAE) controls the bias–variance trade-off through its 𝜆 parameter. However, conventional PPO does not explicitly account for the reliability of individual advantage estimates. In this work, we propose a TDerror-based reliability weighting mechanism that adaptively adjusts the contribution of GAEadvantages according to their associated…

A configuration-driven, end-to-end framework for transforming CMS TICL NanoAOD event data into analysis-ready, validated Parquet datasets and fully reproducible machine-learning experiments. The framework standardizes the complete workflow: extracting and flattening detector-level and truth-level information, performing configurable truth matching, applying deterministic filtering and feature preprocessing, validating schemas and data quality, and producing versioned datasets with complete provenance. It also supports repeatable…

Modern particle detectors generate enormous amounts of high-dimensional data by recording the energy deposited by particles traversing the detector. A fundamental chal- lenge is identifying which detector signals originate from the same particle, particularly in the high-density collision environment expected at the High-Luminosity Large Hadron Collider (HL-LHC). This problem can be formulated as a graph-based pattern recognition task, where reconstructed…

SkillRAG is a skill-aware and explainable CV retrieval and ranking system that matches job descriptions with the most relevant candidates using semantic search, skill coverage, and job-specific scoring. It produces a ranked Top-K shortlist along with evidence-based explanations for why each candidate was selected, with future potential to support diversity-aware ranking and candidate grouping.

What if any 2D image on the internet could become a physical classroom learning aid in minutes? Solidify is an Image-to-3D pipeline that bypasses the need for 3D modeling expertise or expensive cloud GPUs, empowering teachers to produce custom educational models from photos at an 80% cost reduction. Technically, the system supercharges a TRELLIS-2 backbone via LoRA fine-tuning on Thingi10K…

Deploying scientific computing workloads on a single FPGA often encounters severe memory and resource bottlenecks, limiting both performance and scalability. To address these constraints, this project explores a heterogeneous architecture based on a cluster of FPGA-based Multi-Processor System-on-Chip (MPSoC) devices to accelerate large-scale matrix multiplication using systolic arrays. The methodology emphasizes efficient workload partitioning across multiple heterogeneous nodes and systematically…

Tadqeeq is a multi-agent AI system designed to automate ZATCA e-invoicing compliance audits for Saudi businesses. The system employs a pipeline of specialized AI agents that collaboratively extract and structure key invoice information, evaluate it against deterministic ZATCA compliance rules, and generate clear, regulation-grounded explanations of the findings. To enhance reliability and transparency, each explanation is cross-checked against the corresponding…

Tawazun is an energy grid simulation initiative designed to identify the optimal balance between environmental sustainability and economic fairness in Saudi Arabia. The project models the Kingdom's grid to determine the ideal mix of conventional and renewable energy sources that reduces carbon emissions while maintaining affordable electricity prices and supporting energy producers. Additionally, the project examines the impacts of carbon…

The metallicity ([Fe/H]) of stars in the Milky Way provides a fossil record of its assembly, and the ancient RR Lyrae variables are among its best tracers. Spectroscopic metallicities exist for only a small fraction of these stars, but since their pulsational properties depend on iron abundance, the light curves collected by photometric surveys offer a path to metallicity at…

We developed an inference pipeline for Type Ia supernova light curves using realistic Rubin Observatory simulations within the Cosmographi framework. The project forward-simulates sparse, noisy, multi-band photometry; infers SALT2 parameters—stretch, colour, peak time, and redshift—using grid search and machine-learning methods; and uses Bayesian MCMC to constrain shared corrections to the underlying spectral energy distribution. This provides an initial step toward…

Machine learning models usually give one answer with no sense of how confident they should be. Bayesian methods fix this by averaging over many plausible versions of the model, but the standard sampling algorithms misbehave when data is fed in small batches, which is how modern machine learning is trained. This project evaluates the Ray Tracing Sampler (Behroozi 2025), a…

Particle detectors such as the CMS are not perfectly uniform systems and can contain gaps, cracks, or faulty modules. These irregular regions can make pattern recognition techniques less efficient. Clustering techniques such as CLUEstering are suitable for these tasks because they can be robust to irregular and noisy data. However, detector data can contain many features, which can make clustering…

This project investigates when one Large Language Model (LLM) should trust the critique of another LLM. Using GSM8K mathematical reasoning problems, we developed a Solver–Critic framework with Qwen2.5-3B-Instruct and Llama-3.2-3B-Instruct operating in both directions. The Critic first solves each problem independently, then evaluates the Solver's final answer, and revision is triggered only when disagreement occurs. We extract only pre-revision features…

Multiple-choice benchmarks are the backbone of language model evaluation because they are cheap and exactly gradable, yet frontier models now crowd 90% on them, and the standard remedy, authoring harder items, is expensive, slow, and restarts the contamination clock. We propose answer pooling, a transform that makes existing annotations harder instead of writing new ones. Pool the options of N…

Hepatocellular carcinoma is a leading cause of cancer death, and combination anti–PD-1 immune checkpoint blockade with lenvatinib is a standard systemic therapy for advanced disease. However, only ~26% of patients derive durable benefit, and no validated pre-treatment biomarker distinguishes responders from non-responders. Although anti-tumour immunity research has centred on T cells, the myeloid compartment (macrophages and monocytes) is increasingly recognised…

Interpreting the functional impact of genetic variants remains a fundamental challenge in genomics. Functional annotation and phenotype mapping resources represent different channels of evidence on the potential biological impact of genomic variants. These resources are heterogeneous in structure and format, requiring access to multiple independent databases. In this project, we build a pipeline to annotate sets of single-nucleotide polymorphisms (SNPs)…

This project investigates whether astrocyte–neuron communication is associated with the relative resilience of Ex5 excitatory neurons in Alzheimer's disease (AD). Using processed single-nucleus RNA-seq data from BA9, BA7, and BA17 across low, intermediate, and high AD pathology stages, astrocytes and two excitatory neuronal populations—vulnerable Ex2 and resilient Ex5 neurons—were analyzed. CellChat was used to infer ligand–receptor signaling and identify pathways…

Human breast milk contains a living bacterial community whose composition is not fixed. This project investigated whether that composition changes systematically across the postpartum period, and whether such change reflects a biologically coherent shift consistent with retrograde inoculation, the transfer of bacteria from an infant's developing oral microbiome back into milk during nursing. Using 16S rRNA sequencing data from 540…

a single cell analysis of multiple sclerosis patient compered between CSF vs blood to see the immune cell expansion and genes difference across compartment

A reproduction and re-analysis of a widely cited gut microbiome meta-analysis. The original identified bacterial genera shared across diseases by counting how many studies reached significance — a rule that ignores effect size and cannot tell agreement from contradiction. This project pools effect sizes instead, quantifies between-study disagreement, and tests whether the signal generalizes to unseen cohorts.

This project develops an integrated experimental-computational framework to train machine learning models for predicting microbial growth from genomic and environmental information. The framework combines standardized wet-lab growth characterization, cross-site experimental validation, automated genome annotation, and metabolic pathway feature extraction to generate high-quality genotype-environment-phenotype datasets. By integrating microbial growth measurements with genome-derived metabolic features, the project aims to enable scalable prediction…

Childhood acute lymphoblastic leukemia (cALL) is characterized by the disrupted maturation and clonal expansion of bone marrow lymphoid precursors. While standard bulk sequencing stratifies patients into broad cytogenetic subtypes, it averages out transcriptional signals, masking intra-tumor heterogeneity and minor subclonal populations that drive therapeutic resistance and disease relapse. In this study, we developed an end-to-end single-cell RNA sequencing (scRNA-seq) pipeline…

Computational tools that predict whether an aptamer binds a target protein could accelerate aptamer discovery by shortlisting candidates before experimental testing. To be useful, predictors must work on proteins they have never seen. In this project, we investigated AptaTrans, a widely cited binding predictor reporting a ROC-AUC of 0.921. We re-curated its dataset and evaluated it under different ways of…

This project reanalysed a genome-wide CRISPR knockout screen together with four independent transcriptomic datasets to identify genes whose loss increases tamoxifen efficacy in ER+ breast cancer, and to assess which of those candidates carry the least predicted risk of adding to the musculoskeletal symptoms that drive patients to discontinue endocrine therapy. Individual candidate genes did not replicate across datasets, while…

Blood gene-expression signatures can distinguish active tuberculosis from latent infection, but they are discovered in well-resourced laboratories and deployed on low-cost RT-qPCR cartridges. This project built a selection rule that chooses gene panels for their worst-case performance under simulated instrument failure — assay noise, probe dropout, coarse quantisation, narrow dynamic range — and then applied the same stress-testing to the…

This project investigates candidate marker genes associated with molecular subtypes of breast cancer using single-cell RNA sequencing and independent bulk transcriptomic data. Single-cell RNA-seq data from ER+, HER2+, and triple-negative breast cancer samples were processed using Seurat for quality control, normalization, dimensionality reduction, clustering, and UMAP visualization. Harmony was used for batch correction, while silhouette analysis was applied to evaluate…

Pediatric high-grade glioma (pHGG) is an aggressive childhood malignancy with poor prognosis and limited response to immunotherapy. With recent literature affirming that pediatric gliomas are molecularly distinct from adult glioblastoma, immune mechanisms established in adults may not translate directly to this setting. The pHGG immune microenvironment is dominated by myeloid cells spanning a developmental continuum from resident microglia to recruited…

This project investigates whether myeloid–T-cell communication states can distinguish BCG response in non-muscle-invasive bladder cancer (NMIBC) beyond differences in cell abundance. BCG is a standard immunotherapy for intermediate- and high-risk NMIBC, yet approximately 30–50% of high-risk patients do not achieve a durable response, and validated pre-treatment biomarkers remain limited. Because BCG efficacy depends on antigen presentation and T-cell priming, this…

A tRNA is not uniform. Base-paired stems hold the fold together; unpaired loops carry the recognition and modification chemistry. Yet MitoTIP, the predictor clinicians consult for mitochondrial tRNA variants, is read through one pathogenicity threshold applied identically everywhere. I built a position-level structural atlas of all 22 human mt-tRNAs by parsing the published Mamit-tRNA numbering tables, so every coordinate carries…

Using single-nucleus RNA-seq data from the dorsolateral prefrontal cortex of 17 individuals with major depressive disorder (MDD) and 17 matched controls, this project identifies upper-layer excitatory neurons as showing the strongest and most consistent transcriptional signal associated with MDD among 26 cell-type clusters tested. Pathway enrichment analysis reveals coordinated downregulation of genes involved in glutamate transport, cytoskeletal organization, and mitochondrial…

This project investigates the functional network of the human tRNA-repair enzyme Rlig1. By reanalyzing affinity-enrichment mass spectrometry data using R/Bioconductor, we found that active Rlig1 significantly co-enriches with 19 out of 22 enzymes in the human tRNA repair machinery, extending a two-enzyme observation to a pathway-wide finding, placing Rlig1 as a candidate component of the tRNA repair system.

This project engineers an AI-powered platform that performs deep static analysis on APK files prior to installation, systematically scanning decompiled code to detect embedded malware, obfuscated payloads, and unauthorized data access behaviors. All findings are processed by an AI model that distills complex forensic results into a precise risk score and a clear, plain-language report, empowering non-technical users to make…

A multi-layered malware detection framework deployed at the network edge, combining Cowrie honeypot deception, signature-based detection, machine-learning classification, automated containment, and LLM-assisted malware analysis.

This project focuses on safely emulating APT29-inspired attack techniques in an isolated Active Directory environment using MITRE CALDERA. The attack activities are monitored through Elastic Security, Sysmon, Windows Event Logs, and PowerShell logging. The project includes detection engineering, MITRE ATT&CK mapping, threat hunting using KQL/EQL, incident analysis, and remediation. The goal is to evaluate detection coverage, identify gaps, improve detection…

Build a distributed honeypot platform that collects attack data from multiple nodes and aggregates it into a centralized threat intelligence system. The system will simulate various vulnerable services, log attacker interactions, and process the data to generate attacker profiles, IP reputations, and geolocation insights. Aggregated threat information can be exported as actionable feeds for defensive security teams. The platform also…

Cloud environments can become vulnerable due to security misconfigurations that are difficult to identify manually, especially in large infrastructures. Examples include publicly accessible Amazon S3 buckets, overly permissive IAM roles, internet-exposed databases, or insecure EC2 security group rules. This project proposes a Cloud Security Posture Management (CSPM) platform that automatically scans cloud resources using the AWS SDK, detects security risks…

A Wazuh lab testing whether one shared ruleset can catch exfiltration from two threats that look identical: a trusted insider misusing valid access, and an outsider using phished credentials. One shared detection core, every rule proven only once live-triggered. A custom entropy-and-file-size detector specifically detects steganography, and a risk-scoring engine rolls alerts into a single live per-account score.

An exposure-based attack-path hypothesis engine. It combines exposed assets (hosts, IPs, and services) with known vulnerability data, enumerates the attacker chains supported by that data, and ranks them deterministically—helping security teams identify where to fix things first. An LLM evaluates the plausibility of each already-enumerated chain and provides its rationale, along with AI-generated detection guidance for the blue team.

Recreate the security challenges of real space ground station infrastructure by simulating both the adversary and the defender against a satellite ground-to-space command link. The system implements a formal threat model for the CCSDS Telecommand Space Data Link Protocol and builds a parser for its command frames. The attacker role generates malicious telecommand traffic that reproduces real space-cyber attack techniques…

This project develops a post-quantum secure intrusion detection system for a group of five military UAVs connected to a central control centre. Each UAV trains its intrusion detection model locally using its own network traffic, allowing it to learn patterns of normal and malicious activity without exposing raw data. The UAV then securely sends only the trained model updates to…

QUAROM is a secure data-room platform for storing and sharing sensitive, long-lived documents. It applies client-side encryption and post-quantum cryptography, ensuring that plaintext files, raw file keys, and unencrypted private keys are not exposed to the backend. Access is granted on a need-to-know basis through recipient-specific key envelopes, with additional support for quorum-sealed files that require multiple custodians to approve…

Raqib is an AI-assisted IoT security compliance and risk-assessment platform. Its control design and evidence-collection scope are grounded in two frameworks: the Saudi National Cybersecurity Authority's CGIoT-1:2024 IoT guidelines, formally implemented end-to-end as an 81-guideline compliance module, and the OWASP IoT Top 10, which informs the vulnerability categories the platform's collectors target (default/hard-coded credentials, insecure network services, insecure protocols, lack…

Analysing the dissolution of capillary trapped carbon dioxide in underground anticlines

This project investigates copper-alloy nanoclusters as catalysts for the electrochemical conversion of CO into value-added chemicals. The work focuses on tailoring the composition of the nanoclusters to enhance their catalytic performance and achieve better control over the reaction products.

A manual rotary switch selected which of eight perovskite pixels reached the SourceMeter, so every J–V sweep needed an operator at the bench. I replaced it with eight optically isolated photo-MOSFET relays driven by an Arduino, controlled from the same Python session that runs the instrument. Two earlier designs failed instructively — a CMOS multiplexer whose 70 Ω on-resistance exceeded…

This project investigates the electrochemical reduction of nitrate to ammonia utilizing custom-synthesized Metal-Organic Framework (MOF) based catalysts.

Developing efficient anodic electrocatalysts is essential for reducing the energy consumption of electrochemical hydrogen production, while replacing the sluggish oxygen evolution reaction (OER) with thermodynamically favourable small-molecule oxidation provides a promising route toward more energy-efficient electrolysis. Here, we present a comparative investigation of ZIF-based electrocatalysts for oxygen and urea oxidation reactions (UOR), in which ZIF-8 and ZIF-67 electrodes are systematically…

This project develops a validated, data driven workflow to support the development of hard carbon anodes for sodium ion batteries. The work combines scientific literature mining with 3D modelling of disordered porous carbon structures. A literature extraction pipeline was developed and validated using a manually curated benchmark of 582 experimental records from 60 biochar publications. The final pipeline produced 521…

The project focuses on the design and thermodynamic evaluation of a cascaded industrial waste heat recovery system integrated with thermal energy storage (TES). The system recovers waste heat and utilizes it sequentially according to temperature grade, supplying high-, medium-, and low-temperature process demands. Thermal energy storage is incorporated to store excess thermal energy and provide heat when required. The project…

This project focuses on developing polyphthalocyanine based electrocatalysts for electrochemical CO₂ reduction. The catalysts will be synthesized and evaluated through electrochemical testing and product analysis, followed by structural and sample characterization. The project aims to explore how catalyst composition and structure influence CO₂ conversion and product formation, providing insights into the design of efficient and selective catalysts for sustainable CO₂…

This project investigates La and Tb substitution at the Nd site of NBSCF triple-conducting perovskite oxides, using substitution levels of x = 0, 0.25, 0.50, 0.75, and 1.00. The effects of substitution on crystal structure, hydration behavior, and electrical conductivity were evaluated using XRD/PDF, TGA, HRTEM/FFT, and DC conductivity measurements. Results show composition-dependent structural evolution and demonstrate that A-site substitution…

Transparent wood provides daylight transmission by filling a bleached wood scaffold with a transparent polymer. However, polymer filling can reduce the intrinsic thermal resistance of porous wood. Porous silica may help restore thermal resistance through its confined porous structure. the key question of this work is whether we can achieve a transparent wood window that balances transparency and thermal insulation…

Treating NiOx with UVO allows us to activate its surface so Co-SAM survives against the exposure to precursors solvents

The transition from fossil fuels toward clean and sustainable energy technologies has become a global priority and a key strategic goal for Saudi Arabia. This project investigates a benzothiazole-based Ru(II)–PN³ pincer catalyst, synthesized and characterized for selective dehydrogenation of formic acid. As a liquid organic hydrogen carrier, formic acid offers a promising route for the safe storage, transportation, and release…

This project evaluates self-supported NiFe-LDH electrocatalysts grown directly on nickel foam and iron mesh for the urea oxidation reaction (UOR) under alkaline conditions. Bare nickel foam and NiFe-LDH-modified electrodes were tested in 1 M KOH, with and without 0.33 M urea, using cyclic voltammetry, linear sweep voltammetry, and electrochemical impedance spectroscopy. The electrodes were compared in terms of current-density response,…

This project focuses on testing new liquid electrolyte formulas designed to keep high-performance batteries running safely and reliably in harsh environments. The testing subjects them to severe conditions, including high charge/discharge speeds, extended operating ranges, and extreme heat above 50C°.By evaluating how well these chemical additives prevent battery degradation during rigorous stress tests, this work aims to develop long-lasting energy…

This project investigates the high-frequency behaviour of thin-film superconducting hybrid structures for potential applications in ultra-low-dissipation information processing and quantum technologies. The work combines surface characterisation using atomic force microscopy (AFM) with the study of microwave and ferromagnetic-resonance techniques to understand how superconducting and magnetic layers interact at high frequencies. Particular attention is given to phenomena such as superconducting vortices,…

Hydrogen Mixing and Recovery in Subsurface Storage During Repeated Injection–Withdrawal Cycles

The oxygen evolution reaction (OER) is the rate-limiting half-reaction in alkaline water electrolysis and plays a crucial role in the efficiency of green hydrogen production. The activity and durability of NiFe-based OER electrocatalysts are strongly influenced by their synthesis conditions, particularly the choice of metal precursor salts and post-synthesis thermal treatment. This project aims to systematically investigate how different nickel…

We used four different donor–acceptor combinations to prepare the active layers: PM6:L8BO, PM6:Y12, PCE-10:L8BO, and PCE-10:Y12. We then investigated the thermal stability and photostability of each system. For the photostability measurements, we placed identical devices for each donor–acceptor system under 1-sun illumination, with and without a UV filter, to unravel the effect of UV light at the interface and active-layer…

Investigation of passivation at the n-type interface in tin-lead perovskite solar cells for all-perovskite tandem photovoltaics. Perovskite solar cells (PSCs) have emerged as a promising renewable energy technology because of the rapidly improving power conversion efficiencies (PCEs). Mixed tin-lead (Sn-Pb) halide perovskites have narrow (1.25 eV) bandgaps, making them singularly suitable as bottom-cell absorbers in all-perovskite tandem solar cells. However,…

My project focuses on assessing the material flows and supply chain of rechargeable aluminum batteries, with particular emphasis on aluminum, molybdenum, sulfur, and the Mo₆S₈ Chevrel-phase cathode. I mapped the main material and processing pathways and developed a material flow analysis for the aluminum supply chain, covering bauxite production, alumina refining, primary aluminum production, downstream use, and recycling. I also…

Electrochemical CO₂ reduction offers a promising pathway for converting carbon dioxide into valuable chemicals and fuels using renewable electricity. However, achieving high product selectivity while maintaining efficient catalytic performance remains a major challenge. This project explores metal polyphthalocyanines as tunable electrocatalysts for selective CO₂ electroreduction and demonstrates the quantitative selectivity of Ni-based polyphthalocyanine toward CO production. Through structural characterization, electrochemical…

The project focuses on a python-based modeling and analysis of a sorbent-enhanced reforming (SER) process for hydrogen production. The primary objective is to develop a computational framework to evaluate the coupled reaction–adsorption system and quantify its impact on hydrogen yield, CO2 capture efficiency, and overall process performance. The work will involve formulating a reactor model for a fixed-bed SER system…

This internship focuses on monitoring and optimising a photovoltaic-thermal (PVT) and heat pump system at the North Sydney Olympic Pool. The project involves working with operational data to evaluate system performance, quantify heat and energy flows, and assess different control strategies. You will support analysis of how solar input, system operation, and environmental conditions affect efficiency, and contribute to improving…

This project investigates H2 and CO₂ adsorption in porous carbon materials using molecular dynamics MD) and Grand Canonical Monte Carlo GCMC) simulations using LAMMPS software. The aim is to understand how pore structure, density, pressure, and temperature influence the uptake, supporting the development of efficient materials for carbon capture and long-term storage.

This project investigates electrolyte additives for sodium-ion batteries to stabilize electrode interfaces, reduce irreversible sodium loss, and improve long-term battery performance.

Pulsed-Potential Electrochemical Reduction of Nitrate to Ammonia This project investigated the electrochemical reduction of nitrate (NO₃⁻) to ammonia (NH₃) using Cu₂O nanoparticles as a low-cost catalyst. Calcined and uncalcined Cu₂O electrodes were tested in an alkaline flow cell and characterized using SEM, XRD, and EDS. Ammonia production was confirmed by UV–Vis and ¹H NMR, with maximum Faradaic efficiencies of 80.12%…

This project investigates the spatial and temporal evolution of photogenerated charge carriers in gallium arsenide (GaAs) and cadmium telluride (CdTe) using 4D Scanning Ultrafast Electron Microscopy (4D-SUEM). The acquired time-resolved images are processed and quantitatively analyzed using intensity profiles and two-component exponential fitting to characterize the photo-induced surface dynamics. The results reveal different photo-induced responses in GaAs and CdTe, providing…

using the active learning alrogithm L* and a GAN(graph attetion network) as an equeavlince querry to reverse engineer power grid to mapp them , detect failure and how to deal with them

The project focuses on Lithium Extraction from oilfield brines using electrodialysis technology

Grid-forming (GFM) control enables inverter-based resources to actively regulate voltage and frequency, supporting stable operation in low-inertia and weak-grid systems. While most GFM implementations focus on three-phase systems, single-phase GFM is increasingly relevant for residential-scale distributed energy resources. Among GFM approaches, virtual oscillator control (VOC) offers fast dynamic response and inherent synchronization through oscillator-based voltage formation. This project aims to…

Optimal power flow (OPF) sets the operating point of an electrical grid. Its exact alternating-current (AC) form is nonconvex and too slow for near-real-time use. The linear direct-current (DC) approximation is fast, but it returns operating points that are not physically valid on the AC grid. We reproduce and extend a residual-learning framework that maps a DC-OPF solution to its…

Investigated how the surface morphology of Pulsed Laser Deposition (PLD)-grown Niobium Nitride (NbN) thin films varies with nitrogen partial pressure (0–70 mTorr) and determined which morphological features quantitatively correlate with the superconducting transition temperature (TC).

Niobium nitride (NbN) is a strongly-coupled s-wave superconductor central to many technologies where a high and reproducible transition temperature (𝑇c) is essential. Prior work showed that 𝑇c in pulsed-laser-deposited 𝛿 NbNcan be tuned through nitrogen partial pressure via lattice-strain and carrier-density effects, but did not analyse normal- state resistivity directly as a predictor of 𝑇c. Here we combine published transport…

Organic solar cells offer distinct advantages in flexibility, lightweight design, and semi- transparency, yet poor long-term operational stability remains a barrier to commercial deployment. This project studies the structural and chemical stability of organic solar cells. State-of-the-art organic blend thin films are fabricated via spin-coating from chloroform as a solvent using polymer donors (PM6, D18-Cl) paired with non-fullerene acceptors (Y6,…

This project used the e-Hydrogen Cost Optimizer to assess the levelized cost of hydrogen at three coastal sites in Saudi Arabia across 2025, 2035, and 2050. It also quantified the effect of electrolyser degradation on cost and developed a web-based toolkit to make the optimizer more accessible.

This project investigates whether and how geometric information can be reintroduced into coarse-grained graph neural networks (GNNs) for metal-organic frameworks (MOFs). Starting from a CG2NN baseline that encodes only building-block identity and connectivity, the study systematically ablates three levels of geometric detail: (1) RBF-encoded pairwise metal–linker distances, (2) line-graph bond angles, and (3) global persistent-homology descriptors of the pore space.…

Navigating Thermal Dynamics in CO₂ Hydrogenation to Ethanol over Supported Catalysts

This study used a high-flux solar simulator to induce thermal shock in gold ore (600–800°C, 2.5–7.5 min) as a zero-carbon alternative to furnace-based pretreatment. By analyzing mass loss, XRD/XRF mineralogy, BET surface area, and the Bond Work Index, it shows that solar-driven thermal shock creates internal microcracking that cuts grinding energy by up to ~70% (from 558 to 167 kWh/ton…

Water scarcity in arid regions is increasingly intensified by climate change, population growth, and pressure on limited freshwater resources. This study assesses the climate resilience of urban water systems, focusing on desalination, groundwater, wastewater reuse, and rainwater harvesting. Google Earth Engine (GEE) was used to process and visualize climate and remote-sensing datasets, including NASA GRACE MASCON, TerraClimate, and ERA5-Land, to…

This study investigates the treatment of produced water using different draw solutes

This report compares land-cover NDVI derived from UAV, SkySat, and Sentinel-2 imagery at the King Abdullah Monument site at KAUST. UAV28 and SkySat were acquired on 28 June 2021, and UAV30 and Sentinel-2 on 30 June 2021. The observations were evaluated using an overall mangrove region of interest (ROI), three equal-area mangrove-density ROIs, and land-cover ROIs representing mangroves, seawater, asphalt,…

This project focuses on the design and fabrication of novel membranes for advanced water desalination and treatment. The work includes membrane synthesis, structural characterization, and performance evaluation to investigate membrane properties and separation performance. The project provides practical experience in advanced membrane materials and their applications in water treatment.

The main objective of this project is to validate a reliable analytical technique that will allow the analysis of Contaminants of Emerging Concern (CECs) in marine organisms based on the principle of solid phase extraction (SPE) followed by HPLC-Orbitrap Mass Spectrometry analysis. The effectiveness of the analysis technique is estimated in terms of calibration, extraction recovery, matrix effect, sensitivity, and…

This project aims to evaluate the use of AnMBR-treated municipal wastewater mixed with compost leachate for growing lettuce in a controlled hydroponic system. Four conditions will be compared: tap water control, AnMBR effluent, AnMBR effluent with aeration, and AnMBR effluent with aeration and nitrate supplementation. Plant growth will be evaluated through height, root length, leaf number, leaf area, biomass, and…

This project aims to turn fog into a useful and sustainable source of water by improving the design and surface properties of passive fog collectors. It explores how water droplets can be captured from fog and guided efficiently into a collection system without requiring complex cooling equipment. Conventional collectors, including the Fog Harp, metal mesh, and Raschel mesh, are compared…

The excessive emission of green house gases (GHGs) is a leading cause of global climate change, Methane (CH4) and Nitrous Oxide (N2O), although ranked second and third most abundant GHGs, are significantly more potent than Carbon Dioxide (CO2). Both gases often co-exist in wastewater, microbial growth in wastewater environment suggests the existence of a microorganism capable of metabolizing both gases…

A comparative Life Cycle Assessment (LCA) of the operational phase of Activated Sludge (AS) and Membrane Bioreactor (MBR) technologies for municipal wastewater treatment. Using 1m³ of treated effluent as the functional unit, the assessment quantifies key operational inputs and outputs, including electricity and chemical consumption, and evaluates their associated environmental impacts. The main focus is on the applicability of the…

My project focuses on the development of MOF-on-MOF composite materials for the recovery of valuable gold resources from aqueous solutions. The approach uses UiO-66 as a stable and consistent core structure, with the porphyrinic metal–organic frameworks MOF-525 and MOF-545 grown on its surface. Using UiO-66 as a common core provides a similar structural platform for the two porphyrinic MOFs, allowing…

Anion exchange membranes are usually characterised in solutions containing a single salt, yet the water they treat contains several competing anions at once. This study measures water uptake and Donnan selectivity for three commercial membranes, Fumasep, PiperION and Sustainion, in solutions containing five competing anion pairs at three concentrations, giving 45 measurements. Membrane chemistry was the dominant factor: Sustainion absorbed…

Gold, silver and platinum were photodeposited onto anatase TiO₂ at 1 wt% and screened for photocatalytic hydrogen peroxide production under two dissolved oxygen regimes. Gold performed best.XRD, SEM and UV-Vis DRS showed the support was unmodified and the metals finely dispersed. Comparing the regimes showed that oxygen supply, not charge separation, sets the rate for the weaker catalysts.

Combining different types of local bacteria with treated wastewater is a possible strategy for sustainable agriculture. Using a plant growth-promoting bacteria approach has the potential to support plant growth and mitigate the stress that comes naturally with treated wastewater, such as drugs, heavy metals, and excessive nutrient uptake

With Saudi Arabia facing a water crisis and the need for sustainable desalination practices arising, Membrane Capacitive Deionization (MCDI) proves to be an attractive method. MCDI acts as a sustainable electrochemical approach for the desalination of brackish groundwater. With the addition of photovoltaic (PV) cells, it serves as a low-energy solution for remote communities within Saudi Arabia. This project provides…

Sampling of copepod communities from the Red Sea to establish and maintain stable laboratory cultures for use in ecotoxicological testing.

Our project uses AI to convert 2D images into 3D models. We combine neural rendering and photogrammetry tools like COLMAP and OpenSplat to generate realistic 3D outputs from single-view images. This is aimed at enhancing e-commerce, cultural heritage preservation, and digital experiences in initiatives like NEOM and Saudi Vision 2030.

Commercial and residential buildings consume a significant share of global electricity.Traditional rule-based HVAC schedules are static and seldom optimise for varyingoccupancy or weather. We aim to learn an adaptive control policy with Deep Reinforcement Learning (L) that minimises energy use while keeping indoor comfortwithin bounds. Our system should quickly adapt to unseen buildings or climates by transferring experiencelearned elsewhere.

students will work at the intersection of data science, policy, and food systems to analyze critical agricultural research shaping the science of food, agriculture, and the environment in the United States. Using natural language processing and machine learning, the team will explore federal grant records to uncover how agricultural research has adessed issues like food access, food systems innovations, and…

An AI-powered app that analyzes food labels to help users understand nutritional value, ingredient safety, and potential health risks.

This project uses reinforcement learning to discover counterexamples in discrete mathematics, the aim is to automate the search process and assist in identifying flaws in mathematical conjectures.

A team of students led by Ph.D. student Yu Wei and assistant Tong Qiu from the Spatial Ecology and Environmental Data Sciences (SEEDS) lab will utilize cutting-edge remote sensing technologies—including hyperspectral imagery and airborne Light Detection and Ranging (LiDAR)—combined with an advanced deep learning framework to enhance forest biodiversity monitoring. The students will identify key features beneficial for species identification,…

Chatbot which uses camera and microphone input to describe surroundings, read signs, and respond to commands.

The built environment is sensitive to global warming and climate change, which are leading to increased cooling loads, dangerous heat waves, damaging flooding in cities, permafrost degradation, and other impacts. To model, quantify, and predict these impacts for engineering analysis requires “downscaling”, which maps available climate information (e.g. weather station data, model output) to the requirements of engineering (site specific…

Development of a prototype that can classify mosquitoes using acoustic signals. The project will be in conjunction with the 3D printing lab as the TinyML device will work inside a mosquito trap that will be 3D printed.

Cerebral hemodynamics are critically important for diagnosing and monitoring brain diseases. For this purpose, the potential for resting-state functional magnetic resonance imaging (rs-fMRI) has been explored. rs-fMRI a powerful tool previously used for mapping neural activity, is available in most hospitals. Typically, carbon-dioxide (CO2) modulations and recordings are needed in a deconvolution to obtain hemodynamic metrics such as cerebrovascular reactivity…

To develop and test a software tool that generates 3D models (for 3D printers) and/or 2D vector awings (for laser cutter production) starting from photos of buildings and monuments, as a cost-effective inclusive solution to facilitate the promotion of cultural heritage to visually-impaired people.

Detecting the trajectory of the soccer ball and goal keeper with their pose to predict whether the shot would cause a goal. From the collected data we can get statistics such as “Expected Goal on Target (xGOT)”. Furthermore, you can use similarity search from the detections to find similar penalty shots.

We developed an AI-powered ball action spotting system that detects and classifies 12 types of ball-related actions from broadcast football videos, pinpointing the exact timestamps of each event. Our model processes dense, untrimmed match footage to retrieve moments with specific semantic meaning — enabling faster highlight generation, real-time analytics, and advanced coaching insights without manual tagging.

AI-powered code assistants are widely used to generate code completions, significantly boosting developer productivity. However, these tools typically present suggestions without explaining their rationale, leaving their decision-making process inscrutable. This opacity hinders developers' ability to critically evaluate the output, form accurate mental models, and build calibrated trust in the system. To adess this, we introduce COPILOTLENS, a novel interactive framework…

We are constantly exposed to various sensory stimuli such as sight, sound, and smell. Although sensory systems detect and interpret these stimuli, our perception is influenced by internal states such as hunger, stress, and inflammation. However, it is still unclear how the signals that signify these states, such as hormones, peptide, and cytokines, are encoded in gene expression of individual…

Building tools to analyze web scraping at scale based on analysis of Duke's web logs. Investigating the time-scale of AI data scraping and influence of different scrapers by planting content “honeypots” online. Using machine learning techniques like clustering and classification to identify spoofing bots.

Recognize sign language gestures as queries and respond. Possibly in Arabic.

This project proposes a content-iven pipeline for analyzing surveillance videos by segmenting long streams into semantically meaningful, action-centered clips. Leveraging motion cues and object-level understanding, the system enables efficient indexing and retrieval of relevant events using natural-language queries, supporting intelligent surveillance analysis.

Coral Guardian+ is an AI-based system that uses deep learning to classify coral health (healthy, bleached, or dead) from underwater images. It integrates environmental data and temporal trends to improve accuracy and predict bleaching risks.

A team of students will collaborate with a doctoral fellow in Computational Humanities and the Asian American Studies Librarian to develop a statistics-based Natural Language Processing toolkit to study the linguistic styles of Asian American short stories published between 1974 and 2024. This toolkit will support historians, literary scholars, and humanities researchers in general who wish to conduct computational or…

Using Electric Health Record (EHR) data sources for public health analysis. We are currently studying dementia population characteristics using an EHR databases called Truveta.

The application of deep learning to Alzheimer's disease (AD) research using MRI is a rapidly evolving field, with existing studies serving primarily as proof of concept. This Data+ project aims to contribute to the development of a deep learning model that integrates MRI-based topological biomarkers for the early detection of AD. Led by Tananun Songdechakraiwut (Computer Science) and Michael Lutz…

to model mathematical objects from the IMAGINARY collection (and/or from other similar projects) in a way that makes them 3D-printable with low-cost FFF 3D-printers, for educational use in STEM disciplines.

This project aims to revolutionize psychological assessment by leveraging AI to design more reliable and valid psychological scales. By employing machine learning algorithms and natural language processing, existing scales will be analyzed to identify limitations and develop enhanced tools that more accurately measure psychological constructs.As a participant, I will engage in a case study exploring how AI can be of…

References: an article explaining the concept (MIT): https://dspace.mit.edu/bitstream/hand… and a video (YT “Plasma Channel”) on building it: https://www.youtube.com/watch?v=G2brxBRnRH4 Project objective: to design and build a prototype of such a device (ideally suited for possible use in desert environment), to test the feasibility of the process inside a controlled enclosure (with known humidity) and to measure its efficiency (energy used /…

This project develops an intelligent pipeline to fact-check spoken claims in Arabic medical videos, with a focus on diabetes-related content. The system automatically extracts audio, transcribes speech, and identifies factual claims. These claims are then verified using (web search and/or RAG) and large language models (LLMs), enabling real-time assessment of medical information based on trusted online sources.

By leveraging DHM, automated, high-precision and cost-effective diagnostic solutions can be developed for diseases affecting millions worldwide, enhancing healthcare accessibility in resource-limited settings. This project proposes the design, development, fabrication, testing, and validation of a compact, cost-effective, low-cost-per-test, field deployable DHM coupled to a trained ML model for remote label-free automated point-of-care disease diagnosis. The features obtained from DHM can…

In this project, we harness recent breakthroughs in brain‑signal decoding and generative diffusion models to translate raw electroencephalography (EEG) recordings into photorealistic images. Using the newly released Alljoined1 dataset—currently the largest publicly available collection of time‑locked EEG responses to natural images—we aim to build a reliable “thought‑to‑image” pipeline that scales beyond the limitations of current models.

This project adesses the energy crisis in Cape Town, South Africa, where frequent power outages and infrastructure failures have led to inequitable energy access. Wealthier residents can afford solar installations, while low-income communities are left behind. My team and I aim to develop deep learning models to detect and classify different types of solar energy systems (solar panels, solar water…

This project focuses on building an Advanced iver-Assistance Systems (ADAS) perception module specifically for Saudi Arabia's challenging weather conditions (dust storms, sun glare). It involves image enhancement, semantic segmentation, and lane-keeping.

We aim to advance precision oncology by developing robust models to predict ug–target binding affinity (DTA), a key step for tailoring cancer treatments to individual patients. Our project explores graph-based deep learning methods to represent ugs and target proteins, focusing particularly on improving the protein branch of these models.We plan to mitigate information leakage across datasets to ensure reliable performance…

Car type and model can be identified by acoustic fingerprints. The project is about coupling audio-based AI model for car detection with CO2 sensors.

We want to trace how the discourse of commercial consumption and the labor needed to sustain Early Modern Markets is presented 17th century documents produced for the English Trading Companies and responding to the economic exploitation of the New World. An ever-growing supply of new luxury goods reached England through the recently-opened trade routes to the New World, the African…

Large language models (LLMs) have opened up new frontiers for reducing administrative burdens in health systems. Healthcare institutions around the world have already begun piloting the use of automated scribes and other tools aimed at summarizing patients' clinical records. Aside from these institutional endeavors, there is also evidence that independent care providers are increasingly utilizing large-language models to support care…

This project aims to enhance the sports viewing experience by identifying and extracting key moments from match videos. The system detects six events goals, penalties, fouls, shots on target, red and yellow cards, and direct free kicks. For each detected event, it retrieves short highlight clips, allowing viewers to instantly rewatch pivotal moments.Alongside these clips, it integrates a dynamic tracking…

A team of researchers associated with the Applied Machine Learning Laboratory will lead a team of students in developing novel machine learning techniques that will be used for improving brain computer interfaces (BCIs) using electroencephalography (EEG) data. Students will learn how to pre-process EEG data, extract EEG features, and train machine learning algorithms for character selection in a spelling interface…

A group of students, guided by climate science and environmental engineering professors, will use deep learning models to enhance flash flood predictions in the Southeastern United States. They will study extreme weather events that contribute to flooding and learn to identify these events using satellite and radar imagery. By applying their findings, the students will develop and refine deep learning…

This project explores the transition from traditional molecular dynamics simulations to modern machine learning approaches in modeling aqueous systems. By reviewing key literature and engaging with group members, the goal is to synthesize classical physics insights with contemporary data-iven methods, identifying opportunities for hybrid strategies in future research.

AI aided education and language training to help students prepare for the IELTS exam

This project aims to explore the potential of large language models (LLMs) to adess critical challenges in smart grids, such as cyberattack detection and energy forecasting. LLMs excel not only in accuracy but also in generating explainable insights, making them valuable for complex decision-making in energy systems. The project will focus on developing LLM-based frameworks tailored to smart grid applications,…

Why don't more households invest in the stock market? Is it too difficult to open a brokerage account? While this may have been true in the past, advancements in FinTech have made the process simple and accessible, often requiring just a few taps on a smartphone. Instead, could the real issue be that households are simply misinformed about the risks…

The development an equity dashboard in hospitals has been proposed as a solution to facilitate the identification of variations in outcomes, encourage accountability, and support ongoing monitoring. Our research sought to develop an equity dashboard using data collected from the maternal care wards of a hospital in the US in 2019 and 2020. The data obtained were cleaned, and patient…

Summary: If the students are interested in pursuing this project, I will lend them a Raspberry pi along with its camera. The camera will feed the image into the edge device in real-time to run it through an Object Detection (OD) DNN model. Students can interactively show their system during posters session. Applications: security survielance cameras (intruders detection), real-time people…

Summary:If the students are interested in pursuing this project, I will lend them a Raspberry pi. The microphone will feed the voice into the raspberry pi in real-time so it will convert the voice into words. The words will be processed on Raspberry pi using a quantized Llama 3 model. The model's output can be then reconverted into voice.Students can…

Mental health disorders often co-occur, a phenomenon known as comorbidity, posing significant challenges to accurate diagnosis and personalized treatment. Traditional machine learning approaches for comorbidity prediction frequently operate as black-box models, limiting their clinical applicability due to a lack of interpretability. This project aims to adess this gap by developing an interpretable multitask learning framework using Kolmogorov–Arnold Networks (KANs) for…

The objective of this project is to analyze customer interactions by detecting and tracking both individuals and products within Jarir stores. Through advanced detection and analytical techniques, the system aims to provide insights into customer behavior and engagement patterns.

Collaborative project between KAUST Academy and Jarir Bookstore to develop an AI-iven recommendation system tailored for the needs of small- and medium-sized business clients.

The Jarir-NLP project introduces an AI-powered virtual assistant that helps online shoppers navigate Jarir's wide selection of computers and tablets. By simulating the role of an in-store salesperson, the assistant engages customers in natural conversation to recommend devices based on their specific needs, preferences, and budget.Built on a Agentic AI framework, the system combines real-time search with language generation to…

Imaging mass cytometry enables high-dimensional protein marker detection, providing insight to the spatial and cellular heterogeneity of human tissues. Recent advances in 5 micron resolution permit high throughput tissue processing, however, traditional cell segmentation methods fail to accurately segment lower resolution IMC images. To adess this limitation, we propose a novel computational framework that implements pixel-level cell classification for segmentation-independent…

Supervisor Fralick has developed a framework of six domains of study design that can affect the internal validity of randomized controlled trials (RCTs), encapsulated by the acronym PHOBIA: Placebo controlled? How was it funded? Outcome clinically valid? Blinded? Intention-to-Treat? A lot of centres and patients included? When evaluating an RCT, these 6 elements are crucial considerations. The current paradigm leaves…

There are several mesh architectures based on LoRa but there are no studies thatcompare them scientifically. The project is about establishing a testbed to comparedifferent LoRa mesh solutions.

SoC-FPGA technology is increasingly used for hardware acceleration of ML models thanks to its key advantages, including high parallelism, low power consumption, and low latency. However, mapping an ML architecture to a SoC-FPGA often requires the application of compression strategies to create a hardware design that effectively balances model performance, memory footprint, and inference time. This project focuses on employing…

The project develops and benchmarks advanced deep-learning modules that automatically clean and reconstruct the multi-terabyte timestreams recorded by the Simons Observatory and its precursor, the Atacama Cosmology Telescope. By suppressing instrumental noise and classifying foreground galaxies and stars, these models generate high-fidelity CMB maps while flagging potential astrophysical transients such as flaring stars and gamma-ray bursts. The work integrates convolutional,…

An AI to help by scanning messages, detecting blackmail words, and tracing the IP adesses where the threats come from.

CMOS image sensors, initially designed for visible light imaging, are also sensitive to ionizing radiation, making them suitable for use as radiation detectors. Key advantages ofthese sensors include their low cost, ease of readout, ability to operate at room temperature, capability to perform X-ray spectral measurements, and high spatial resolution due to their small pixel pitch. This project focuses on…

Distilling and fine tuning a VLM to achieve SoTA metrics on certain benchmark datasets in the area of surveillance VQA

A Retrieval system that retrieves past foul verdicts to help referees make a decision on a similar present one

This project explores numerical literacy across generations in the United States using the Panel Study of Income Dynamics (PSID), a comprehensive longitudinal dataset. Due to the dataset's size and complexity, we use AI to optimize performance and processing. The study examines how family structure, income shifts, and changes in socioeconomic status, such as becoming richer or poorer, are linked to…

An Open-Source Tool to Analyze Longitudinal Data without Geographic Harmonization

This project aims to develop a corporate AI agent that functions as a full-time digital HR assistant. In response to a real-world use case from Elm, we are building the Policy and Procedures Agent, designed to provide 24/7 support on HR policies.The agent will:- Answer policy-related questions in Arabic and English.- Tailor responses based on employee location, grade, or business…

This Data+ project aims to improve microbial growth prediction using custom analytics, enhancing the MicroNet web app for better data visualization and statistical modeling. Students will beta-test the tool, extend its models, and analyze data to advance applications in health, food safety, and microbiome research.

We would like to share our proposed project idea: Real-Time Analytics for sports matches, focusing on live tracking and visual insights .

This project presents a real-time video interpretation system that captures live camera feeds, detects objects using YOLO, and generates natural language captions for key frames using multimodal models. These frame-wise captions are then aggregated and summarized every minute using large language models (LLMs), such as Cohere's command-r-plus, to produce a concise human-like scene description. The pipeline includes intelligent object detection,…

Developing computationally intensive scientific algorithms by leveraging the complementary strengths of different processing units (FPGA, CPU, and GPU) within MPSoC devices, optimizing performance, flexibility, and energy efficiency across diverse application domains.

Implementation and optimization of Sparse Matrix-Vector Multiplication (SpMV) on heterogeneous MPSoC, utilizing FPGA for sparse data indexing and memory-efficient operations, GPU for parallel computation of nonzero elements, and CPU for workload distribution and task scheduling. The proposed solution targets HPC applications such as scientific simulations, machine learning, and graph analytics, optimizing execution speed and energy efficiency in large-scale sparse computations.

TinyML, Mosquito Classification based on Aquatic signals on microcontrollers

TinySecure Vision is an advanced embedded security system designed for real-time face detection and recognition, using optimized AI models for microcontrollers and lightweight protocols to ensure fast and reliable access control and monitoring.

A team of students led by Hubbard Brook Experimental Forest researchers will develop a machine learning approach to identify aquatic insects and contribute to a long-term ecological dataset being used to examine changing insect population dynamics. Students will use a set of existing images to train a supervised machine learning model to classify images of insects and, using additional climate,…

Social determinants of health are increasingly acknowledged as key factors in achieving equitable, efficient, and patient-centered care. It is now well recognized that factors such as inadequate transportation, economic hardship, language barriers, employment security, and health literacy play a critical role in patient's care experiences and health outcomes. To this effect, understanding the prevalence of these determinants and their impact…

The recent advances in low-cost radar technologies coupled with TinyML allows the development of a mosquito counting prototype.

Cerebrospinal fluid (CSF) flow helps wash metabolic waste from the brain, which is directly related to brain health, and is known to slow with aging and in Alzheimer's disease. Disruptions in CSF flow have been linked to impaired clearance of neurotoxins such as beta-amyloid, which build up in Alzheimer's disease and may play a key role in early disease progression.…

YouTube is the world's top destination for self-learning, yet it lacks the interactive structure needed for real understanding — users watch but rarely implement. Unlike structured courses that combine theory with hands-on practice through lab quizzes and notebooks, YouTube leaves learners passive. This idea proposes an AI-powered system that converts educational YouTube videos into Coursera-style Jupyter notebooks, complete with a…

We're attempting to adjust the architecture of current state-of-the-art WiFi-based Human Activity Recognition models by fusing a pre-trained Visual Transformer with a diffusion model in search of improved performance.

Algae and aquatic plants are promising biofuel sources because they don't need farmland or freshwater and contain both sugars and oils. Unlike food crops, they avoid the food-vs-fuel conflict. But their biomass is hard to break down due to tough cell walls made of cellulose, lignin, and complex polysaccharides.Current enzymatic methods are slow and expensive — enzymes like cellulases struggle…

This project aims to develop an AI-iven platform for the design of novel peptide adjuvants that enhance vaccine efficacy. By leveraging publicly available immunological data and machine learning techniques, the platform seeks to reduce development costs by over 80% and improve vaccine accessibility, particularly in low-resource and underserved regions.

This project estimates how many coronavirus disease 2019 (COVID-19) infections were missed by relying on deaths rather than reported cases. I combine age-specific infection fatality risk (IFR) estimates from Ward et al. with each country's age structure to create age-standardized, era-specific IFRs.Weekly case fatality rate (CFR) values are computed from reported deaths and cases (no lag). I then infer a…

This project investigated the exchange of gut microbiota between young and aged mice through co-housing. By analyzing microbial composition at the genus level, we observed that specific taxa were transferred in both directions (young-to-aged and aged-to-young), including beneficial bacteria and taxa with context-dependent effects. These findings highlight how co-housing accelerates age-linked microbiota exchange and may provide insights into microbiome-based strategies…

Looking into two RNA-Seq datasets for the C. Elegans to find if there is a relationship between the stages of embryo development found in the bulk dataset. And clusters of cells categorized by time and anatomy found in the Single Cell dataset

The research aims to gain a deeper understanding of the stages of nervous system development in C. elegans through protein expression analysis and to elucidate the differences between the first and fourth stages. Proteins were normalized, clustered, and examined for differential expression. Gene Ontology enrichment revealed that nervous system biological processes are highly enriched at L1, consistent with the early…

This project investigated the transmission of drug-resistant tuberculosis between Botswana and Namibia using whole-genome sequencing. By analyzing genomic data from over 300 bacterial isolates, we identified clear evidence of cross-border transmission events involving multidrug-resistant strains. The key biological finding is that while drug resistance has emerged multiple times independently (polyphyletic), it is also potentially spread through cross-border chains.

Investigate association of epigenetic age acceleration (epigenetic age – chronological age) with cystic fibrosis severity including CFTR genotype, measures of pancreatic function, age of chronic infection with Pseudomonas aeruginosa, age of diabetes onset, sweat chloride levels and lung function.

Variation in gene expression underpins variation in organismal traits and diversity. Therefore, understanding how gene expression evolves will allow us to better understand the mechanisms of evolutionary change. The strength and form of selection on gene expression and its role in evolution is difficult to estimate, however, because of the high dimensional and highly correlated nature of gene expression data.…

Assisted reproductive technology (ART) refers to any fertility treatment in which oocytes (eggs) or embryos (fertilized eggs) are manipulated in a laboratory. The first step of ART is controlled ovarian stimulation (COS), which involves daily injections that stimulate the growth of multiple ovarian follicles. Eggs are then retrieved from these follicles. Predicting a patient's response to COS is challenging. Ovarian…

Embryogenesis in C. elegans is driven by stage-specific changes in gene expression over time. This project aims to explore how transcriptional programs shift during development using time-course bulk RNA-seq. Through differential expression testing and clustering, we identified over 8,000 genes with dynamic expression patterns. Gene Ontology analysis showed that each gene cluster is enriched in distinct biological processes—such as cell…

Multi-omics approach integrating proteomics with single-cell gene expression

Electron microscopy (EM) enables detailed visualization of cellular and subcellular structures, such as mitochonia and nuclei, providing critical insights into biology. However, manually segmenting and annotating large-scale EM datasets is time-intensive and challenging. AI offers a solution by automating this process, but successful model training requires high-quality, manually labeled images to establish ground truth. This project aims to generate labeled…

Widespread deposition of extracellular amyloid-β (Aβ) plaques are the neuropathologic hallmarks of Alzheimer's disease . Studying different cells surrounding Aβ plaques can reveal molecular pathways driving their reactive states and provide insights into how histopathological hallmarks contribute to neurodegeneration. The aim of this project is to understand the cells interactions and whether specific genes are differently expressed near Aβ plaques.

People spend nearly 90% of their time indoors, where they are exposed to various airborne contaminants. Indoor air quality (IAQ) has a substantial impact on human health and comfort. However, understanding and analyzing IAQ in diverse indoor environments remains challenging despite the well-established principles of mass transfer and fluid dynamics and various low-cost sensing technologies. This is due to the…

This project aims to identify differentially expressed genes shared between the progressive forms of Multiple Sclerosis (MS); Primary Progressive MS (PPMS) and Secondary Progressive MS (SPMS) to better understand the molecular mechanisms driving disease progression. A bulk RNA-seq transcriptomic analysis was conducted using publicly available brain tissue data (GEO accession: GSE290552). The data were analyzed using DESeq2, with pairwise comparisons…

This study aims to investigate how aging influences the composition and diversity of the gut microbiome in mice through an integrative , multi-dimensional analysis. By comparing microbial communities across different age groups. The project will identify specific taxa and microbial patterns associated with aging.

We are conducting a literature-based network analysis to investigate key biomarkers involved in mechanotransduction, transport, and angiogenesis in syncytiotrophoblast cells, with a particular focus on the BeWo cell line. Studies published between 2015 and 2025 were reviewed to compile a comprehensive list of relevant biomarkers. A pathway enrichment analysis is currently being performed to identify shared biological pathways associated with…

This project will identify Variable Number Tandem Repeats (VNTRs) that overlap protein-coding regions of the genome, with a focus on a region of chromosome 5 linked to cystic fibrosis (CF) lung disease. Using the adotto repeat catalog (v0.0.2) and the MANE release 1.0 (Matched Annotation from NCBI and EMBL-EBI) database, we aim to find VNTRs that may disrupt splicing or…

Its a single rna sequence from postpartum brain tissue, i uses R studio to analyze the data of 95 thousands of cells total of autism and control

This study investigates protein expression dynamics in Caenorhabditis elegans during larval development, focusing on early (L1) and late (L4) stages. Proteomic analysis identified 244 proteins with significant expression changes out of 4,130 analyzed, revealing distinct patterns of upregulation and downregulation. Bioinformatics analyses, including clustering and protein interaction networks, highlighted stage-specific proteins: ACN-1, which peaks at L4 and is linked to…

This research explores how puberty influences gene expression in children with Autism Spectrum Disorder (ASD), with a focus on the roles of steroid hormones and immune pathways. Using RNA-seq expression data (GSE221645) from whole blood samples, the study compares ASD children to their unaffected siblings both before and after puberty.

Looking through S100 genes family for an immune therapy target to Glioblastoma

This project involved a comprehensive bioinformatic re-analysis of a public proteomics dataset to characterize protein expression dynamics across C. elegans larval stages (L1-L4). Key findings include the identification of stage-specific protein markers, with L1 enriched for metabolic processes and L4 for reproductive functions, and the discovery of gradual temporal expression trends. The study provides a high-resolution proteomic map and a…

This project explored proteomic differences between the proliferative (C1) and differentiated (C4) subtypes of high-grade serous ovarian carcinoma using CPTAC datasets. By applying PCA, heatmaps, and differential expression analysis with limma, we identified subtype-specific proteins and pathways. Enrichment analysis revealed that C1 tumors are enriched in ribosomal and DNA replication proteins, while C4 tumors show detoxification and immune-related processes. Finally,…

This project investigated alternative splicing isoforms as candidate biomarkers for Major Depressive Disorder (MDD). Peripheral blood RNA from MDD patients and controls was analyzed using the nf-core/rnasplice pipeline, with splicing events quantified by rMATS v2.10. Isoforms were first filtered based on false discovery rate (FDR) and percent spliced-in (PSI), using read counts only as a minimum threshold to ensure reliability.…

This project focuses on advancing the STINGAR security platform by designing and deploying new honeypot types using rapid prototyping techniques powered by large language models (LLMs). These updates will improve the detection and mitigation of cyber threats, strengthening the protection of Duke's network and those of over 70 partner universities nationwide. Students will collaborate with Duke's Office of Information Technology…

This project focuses on creating a front-end application for APIARY, a system designed to manage blocklists and safelists for security operations. While the backend API is fully developed, the system needs an intuitive interface to streamline management tasks like adding entries, viewing logs, and performing essential security operations. Students will work with Duke's Office of Information Technology (OIT, IT Security…

This project focuses on advancing the MISTRAL platform, a National Science Foundation (NSF)-funded initiative designed to capture and analyze network data in research lab environments. Students will collaborate with the MISTRAL team, Duke's IT Security Office (ITSO), and key stakeholders to improve tools for querying, distributing, and visualizing security logs and research data. The team will work on analyzing network…

This project focuses on creating an engaging, interactive experience to raise cybersecurity awareness among students. Using gamification techniques, the team will design and develop a hands-on activity, such as a security-themed escape room, capture-the-flag challenge, or phishing simulation. These activities will immerse participants in solving real-world cybersecurity problems, like decoding encrypted messages, spotting phishing attempts, or configuring secure systems, fostering…

This project focuses on exploring machine learning (ML) techniques to enhance the efficiency of the Security Operations Center (SOC). By automating lower-level alert triaging, the goal is to save time and reallocate resources to higher-value tasks. Students will design and build tools that leverage ML for data preprocessing, feature optimization, and advanced alert analysis. The project will also include developing…

This project focuses on upgrading Duke's Black Hole Routing (BHR) system, a key tool for blocking malicious IP addresses and protecting network security. The current system relies on outdated architecture and interfaces, limiting its performance. Students will modernize BHR by developing a portable, containerized solution that integrates with modern infrastructure and can be shared as an open-source tool with other…

This project aims to harness atmospheric water from humidity through the principle of radiative cooling, utilizing various repurposed waste materials, primarily plastics. The entire system is designed to be constructed from a diverse range of recycled and upcycled components, emphasizing a sustainable approach to resource management. Furthermore, this innovative system is engineered for remote, passive operation, meaning it is entirely…

I am currently conducting research at the KAUST Catalysis Center at King Abdullah University of Science and Technology (KAUST), under the supervision of Huabin Zhang and in collaboration with PhD student Abdulrahman Allangawi, as part of a broader initiative supported by Aramco. The focus of my project is the synthesis and development of g-C₃N₄ (graphitic carbon nitride) — a highly…

Project focuses on developing and evaluating novel catalysts for the electrochemical reduction of carbon dioxide (CO₂) into value-added products. Various materials are synthesized and characterized to understand their structure–activity relationships, followed by electrocatalytic testing under different operating conditions. The goal is to identify efficient, selective, and stable catalysts that can enable sustainable CO₂ conversion, contributing to carbon-neutral energy solutions

Comparing two approches in preparing bi-cu, which one is the best for Co2 reduction.

Chiral organic molecules adsorbed on top of a superconductor. These chiral molecules have been proposed to convert a spin-zero Cooper pair to a spin-one Cooper pair thereby allowing spintronics to be integrated with superconductivity. The project will involve understanding how the CISS effect works in the normal state without superconductivity. Once this understanding is developed, the project will explore how…

We are comparing the performance of zinc-ion batteries with zinc hexacyanoferrate cathodes on carbon paper to it on Ti mesh.

This project explores the development of advanced photoelectrocatalytic materials for solar-iven hyogen production and CO₂ reduction, aiming to support Saudi Arabia's Vision 2030 for sustainable energy, under the supervision of Yun Hau Ng and mentorship of s. Rohul Adnan, Naveen M H, and Aliyah Fakim, I am working in the Catalysis for Alternative and Renewable Energy Lab at KAUST.The research…

The project aims to improve zinc-ion batteries by testing and evaluating better cathode materials and enhancing their performance for safer and more efficient energy storage with higher capacity and longer life cycles

This project is an introduction to the study of electrolyte additives for aqueous zinc ion batteries(AZIBs). The AZIB is one of the emerging technologies in the energy storage industry, and the inefficient zinc anode hinders its further application. Electrolyte additives are among the most promising strategies for enhancing zinc anode performance. In the project, we conducted the fabrication and testing…

I investigate stability and self-doping in tin/lead perovskites by studying photoluminescence (PL), PL quantum yield (PLQY), and carrier lifetimes to understand how dopants and processing treatments impact defects and device performance.

This project focuses on synthesizing BaCe₁₋ₓYₓO₃₋δ (x = 0.05, 0.10, 0.15) via solid-state reaction using BaCO₃, CeO₂, and Y₂O₃. The powders will be calcined at 1000 °C, characterized by XRD, SEM, and TGA, then pressed into dense pellets (sintered at 1200 °C). Proton conductivity will be measured between 300–600 °C under y and wet air conditions.

Wide-bandgap (WBG) perovskite solar cells (PSCs) have recently attracted significant interest due to their potential in high-performance photovoltaic applications. In this context, WBG PSCs have shown notable advancements in efficiency and compatibility with multi-layer device structures. To achieve the desired bandgap, bromide is typically incorporated alongside iodide in controlled proportions. However, these mixed-halide compositions often exhibit induced phase segregation, resulting…

Our project focuses on lithium extraction from seawater using advanced adsorbent materials known as Covalent Organic Frameworks (COFs). We synthesized COFs with tailored porosity and chemical functionality to selectively capture lithium ions. The synthesized materials were characterized using Powder X-Ray Diffraction (PXRD) to confirm their crystalline structure, Fourier Transform Infrared Spectroscopy (FTIR) to identify functional groups, and Inductively Coupled Plasma…

Aligned with Saudi Arabia's Vision 2030, renewable energy has become a national priority, placing solar technologies at the forefront. Enhancing solar cell efficiency requires a fundamental understanding of how free carriers, electrons and holes, behave at the surface where most light absorption occurs. Our research utilizes a one of a kind Ultrafast Scanning Electron Microscopy (USEM) to probe charge carrier…

Passive Radiative Cooling for Thermal Management of ones in Harsh Environments

This project focuses on optimizing BiOI thick films for X-ray detection applications. BiOI is a promising material due to its high atomic number and suitable electronic properties. The films are fabricated under varying conditions to enhance their structural and functional performance. Characterization techniques include X-ray Diffraction for crystallinity, a profilometer for thickness and surface profile analysis, and other methods to…

The project is still in its early stages, and some aspects may evolve as I continue exploring the topic and coordinating with my mentor

Thin films are essential in CO₂ applications such as gas separation membranes, reduction catalysts, and sensors. They are tailored for high permeability, selectivity, catalytic activity, or conductivity to enable efficient CO₂ capture, conversion, and detection.

Thin film photovoltaic devices are complex and challenging to make as different materials need to be grown on top of one another in thin film form with minimal defect density. Defects in these thin films can act as trap centres leading to unwanted carrier recombination thereby reducing the efficiency of these devices. In this project we will understand how a…

Development of an efficient electrocatalyst for CO2 reduction into formic acid