Portfolio v2
A high-performance personal portfolio built with Astro, Tailwind v4, and a custom neon cyberpunk aesthetic.
I’m a junior computer science and math double major at the University of Texas at Austin. I’m interested in building intelligent machine learning systems to solve real-world problems in domains such as operating systems, robotics, and quantum computing. Check out my experiences, projects, and blog to take a look at what I’ve been up to.
Skills I've acquired throughout the years.
Selected work showcasing my technical skills.
A high-performance personal portfolio built with Astro, Tailwind v4, and a custom neon cyberpunk aesthetic.
Smart agriculture system using ESP32 sensors and TFlite for disease detection in crops.
WebSocket-based chat application with Redis pub/sub for horizontal scaling across multiple servers.
Fault-tolerant key-value store using Raft consensus algorithm.
Implementation of artistic style transfer using VGG19 network.
Headless e-commerce platform with Stripe integration and inventory management.
Automated code review assistant using GPT-4 for pull request analysis and suggestions.
Progressive Web App for tracking workouts with offline support and data visualization.
Stream processing pipeline for analyzing application logs in real-time using Apache Kafka and ClickHouse.
Git-based content management system with visual markdown editor and version control.
Thoughts, tutorials, and insights on software development.
A presentation I delivered at Polygence's 8th Symposium of Rising Scholars on my quantum computing research. My project was about solving the Max Cut problem using QAOA to find the ground state of Ising spin glasses.
A timeline of my professional journey.
Developing AI-powered tools for Google's internal fuzz testing of several functions in Java codebases using Google ADK.
Integrated SeBS (Serverless Benchmarks Suite) workloads (e.g., video processing, graph algorithms) to improve model accuracy and characterize multi-application interference. Expanding GraphSAGE-based resource contention modeling, which generated 32-dim embeddings from system resource graphs for gzip, bzip2, and xz. Refining heterogeneous graph modeling, addressing previous limitations of homogenized node attributes. Creating an efficient representation of heterogeneous node features to improve contention prediction accuracy. Currently developing a dynamic microservice scheduler through Bayesian optimization and graph application embeddings to outperform standard heuristic-based schedulers like Kubernetes.
Developed a non-autoregressive transformer model for predicting human motion trajectories in hallways. Collected markerless motion capture data on 8 participants in a T-shaped hallway. Processed with C-motion software to extract key body joint information like positions, angles, velocity, and acceleration time derivatives. Preprocessed and tokenized body tracking data with NumPy/Pandas and implemented attention mechanisms to process this continuous state information using PyTorch. Evaluated model performance w/ MPJPE metric and compared against state-of-the-art STPOTR, performing about 50% better with the highest at 0.15 m error compared to STPOTR's highest at 0.3 m.
UGCA for C S 331: Algorithms and Complexity with Dr. Fares Fraij, Fall 2025. Supervise discussion & study sessions of 30+ students, conduct office hours, grading tests and assignments. Answering questions on ED Discussion Forum, creating problems for course tests, proctoring exams.