Autonomous Robotic Foosball
Autonomous foosball-playing robot controlled by closed-loop visual feedback on an Nvidia Jetson. Deploys quantized custom EfficientNet ONNX models via TensorRT to control stepper motors at 30fps.
View Architecture →I'm a recent ECE/CS graduate from Duke University. My academic, internship, and personal projects span software engineering, hardware design, embedded systems, networking, security, and robotics. You can see some of them below!
Note: Repositories private as some projects are still active assignments. Access available upon request.
Autonomous foosball-playing robot controlled by closed-loop visual feedback on an Nvidia Jetson. Deploys quantized custom EfficientNet ONNX models via TensorRT to control stepper motors at 30fps.
View Architecture →Custom 5-stage pipelined processor with MIPS-based ISA implemented on Xilinx FPGA. Features full hazard control (bypassing, flushing) to drive a scratch-coded (Atari) Pong game at 100MHz.
View Architecture →A low-latency trading pipeline PoC: a C++ feed handler on an RP2040 streams orders over SPI to an Artix-7 FPGA implementing an O(1) systolic-array order book and hardware risk checks in Verilog, measuring tick-to-trade latency. Real-time telemetry via UART to a Rust TUI dashboard.
View Architecture →High-throughput trading engine with C execution and Go settlement. Includes custom slab allocation for zero-GC matching, binary framing over Unix Sockets to reduce jitter, and a batch-commit protocol that aggregates database transactions.
View Architecture →End-to-end pipeline that scrapes C repositories and detects vulnerabilities using CodeBERT embeddings. Identified integer overflows and OOB writes in public code and generated concrete fixes via Gemini API.
View Architecture →High-performance implementation of the Quadratic Sieve algorithm for integer factorization. Optimized generation of B-smooth numbers to demonstrate RSA vulnerability.
View Architecture →Naive Bayes sentiment model trained on scraped Twitter data. Built a full ETL pipeline using Snscrape and Scikit-learn; correctly called both 2016 primaries and the general election.
View Architecture →Formal, Assertive and Probabilistic Verification Experiments on Verilog Code
| Core Languages | Python [PyTorch, Flask, Django, Pandas, NumPy], JavaScript [TypeScript, React], Go [gRPC], SQL [PostgreSQL], C++/C, Verilog [Vivado, RTL] |
|---|---|
| Infrastructure | AWS [ECS, Lambda, RDS], Terraform, Docker |
| Databases | PostgreSQL, ElasticSearch, DynamoDB |