Short write-ups with code and demos. Each highlights the problem, the design trade-offs, and the results.
Tauri-based AI research canvas combining a Plan-Execute LLM agent loop, two-stage BGE + RRF retrieval, and an MCP-compatible inspection/automation server.
Open-source Python library (in development) for inspecting and steering robot foundation models — VLA models, world models, RL policies.
Full‑stack stock analytics with live market data, watchlists, interactive charts, and P/L tracking
Built NLP pipelines for sentiment analysis, POS tagging, and NER, progressing from classical ML to neural models with embeddings and transformers
Extended interpreter for a typed functional language with parsing, evaluation, and type inference
Designed and implemented motion planning, localization, and control for a 6-DOF robotic arm in both simulation and real deployment
Scalable pHash indexing with tolerant matching for dedup and archive retrieval.
A compact, well-structured file system + shell I/O layer in C.
Developed a probabilistic genome sequence alignment algorithm improving accuracy and query speed over BLAST for ambiguous genomic regions
Studied recognition–generative pathways of Helmholtz networks with information-theoretic analysis
Improved Unity’s Recast & Detour navigation mesh toolset with analysis of polygon decomposition methods and performance trade-offs
Real-time vision system integrating Kalman filtering, PID control, and terrain-aware path planning for drones
Drone navigation system using vision-based terrain classification and probabilistic path generation
