Things I’ve launched
Surgical robots, GPU clusters, LLM pipelines, and the occasional pothole. Open an entry to watch it run.
LLM assistant · 3D venue viewerLiveLens
A live-event seat-review platform: search venues and seats, rate the view and sound, and explore an interactive Three.js 3D venue — with an LLM assistant that answers questions through 10 database-query tools.
Watch it run
8.9× build speedupMulti-GPU ANN Search
Extended the CAGRA graph ANN algorithm into a multi-GPU vector-search framework: sharded scatter-gather, zero-copy shared-memory loading of SIFT1B, 50M+ vectors indexed at >97% Recall@10.
Open log
11 agent tools · local-firstReview++
An AI study assistant for your own course folder: it reads every PDF, slide deck and note locally, builds a searchable knowledge base, and answers with citations back to the page — plus quizzes, flashcards and mind maps.
Watch it run
EM 22.8% → 69.0%RAG vs. Closed-Book LLMs
A local RAG pipeline pairing 4-bit Llama-3-8B with FAISS retrieval over 21K passages. Benchmarked 4 retrieval strategies across 7 configurations, and built an EM/F1/MRR + LLM-as-a-judge evaluation harness.
Watch it run
50 vehicles, livePotholePatrol
A real-time IoT simulation of crowdsourced pothole detection: 50 vehicles on OSRM-routed GPS paths streaming telemetry to a live map, with a TTL-bounded V2V hazard-sharing protocol and a YOLOv8 vision gate.
Watch it run
8.1× speedup on 10 threadsParallel & Distributed SAT Solver
Serial, multithreaded, and MPI Boolean SAT solvers on a DPLL search with unit propagation — a work-stealing scheduler delivers near-linear scaling, then scales out to distributed memory over MPI.
Watch it run
~9,500 songs modeledSong Popularity Prediction
An end-to-end ML pipeline joining 328K Billboard chart entries with 170K Spotify tracks, benchmarking KNN, SVM, Random Forest, and a PyTorch network — shipped as a CLI that rates any FLAC/MP3.
Watch it run
Real-time occupancyGatekeeper Insights
An embedded customer-traffic monitor in C on BeagleBone + Arduino: motion and laser-photocell sensing, NeoPixel/PRU displays, and UDP streaming of real-time occupancy to a Node.js dashboard.
Watch it runMore on my GitHub.