Federated Learning With Byzantine Clients Tolerance
Building and evaluating a Federated Learning framework using Flower to handle Byzantine (malicious or unreliable) clients in real-world distributed environments.
Research‑driven. System‑oriented. Builder at heart.
Exploring machine learning, distributed computing, and web development to craft elegant solutions.
King Abdullah University of Science and Technology (KAUST)
Math of Machine Learning, Data Analytics, Computer Networks, Concurrency
The University of Texas at Austin
Machine Learning, Virtualization, Cloud Computing, High Performance Computing, Computer Graphics
UT Austin - Longhorn Developers
I contributed as a Frontend Fellow to UT Registration Plus, a student-built Chrome extension used by 50,000+ students to streamline course registration and scheduling at UT. I collaborated with the team through GitHub Issues, PRs, and project boards to improve UX/UI, deliver new features, and prototype a direct course registration flow, including the sign-in process and course data extraction logic.
Aramco - Aramco Research Center
I enhanced the synthetic training data for a computer vision model by creating and refining 3D renders in Blender, automating the workflow with Python, and integrating 360-degree HDRi images, resulting in a large accuracy boost. I also contributed to a change-detection project by preparing datasets, testing models in PyTorch, and building a flexible experimentation repo.
UT Austin - Software Engineering Class
I served as a course TA, mentoring six student teams on full-stack project architecture, holding weekly office hours, and giving detailed feedback on 30+ student blogs while reporting progress to the professor. I also rewrote the auto-grading script with proper error handling, reducing grading time down to about 5 minutes per assignment.
USC Viterbi School of Engineering - Data Science Lab
I implemented the Canonical Polyadic (CP) tensor decomposition algorithm using the Tensor Algebra Compiler and ran experiments on arbitrary datasets, achieving over 90% compression while preserving statistical significance. I also presented the algorithm and results at a department-wide symposium.
Building and evaluating a Federated Learning framework using Flower to handle Byzantine (malicious or unreliable) clients in real-world distributed environments.
Developed GILD, a paid messaging platform enabling user-to-user email communication with integrated balance management and transaction fees tracking.
Participated in a 3-month competition (SDAIA) to enhance ALLaM, a large Arabic language model, focusing on Arabic poetry generation improvements.
Developed a Python-based tool using scikit-learn and CodeCarbon to analyze model accuracy vs. energy consumption during hyperparameter tuning.
Collaborated with a team of 5 to develop a React web application with a MySQL database, showcasing data about wildfires, nearby fire protection facilities, and California counties.
Open for opportunities and collaborations.
Email me at
or use the form below: