CS @ University of Washington Bothell · Building things at the intersection of AI, full-stack, and systems
- Built BridgeLearn — an offline Socratic tutoring system powered by Gemma 4 E4B via Ollama, submitted to the Kaggle Gemma 4 Good Hackathon
- Running local LLMs on consumer hardware — deployed a 4-bit quantized Gemma 4 model with full GPU offloading on a GTX 1050 Ti using WSL2 + Ollama
- Looking for Summer 2026 internships in software engineering / AI / ML
- AI/ML systems & LLM deployment
- Full-stack development (React, Flask, Firebase)
- Systems programming (C++, low-level optimization)
- DevOps & cloud (Docker, Linux, WSL2)
Languages
Frameworks & Tools
| Project | What it does | Stack |
|---|---|---|
| ⚡ CUDA Matrix Multiplier | 3 progressively optimized GPU kernels — naive → tiled → register-blocked — benchmarked vs cuBLAS | CUDA, C++, Python |
| 🤖 Local AI Ecosystem | Deployed a quantized LLM locally with GPU offloading + Streamlit UI + Discord bot | Python, Ollama, WSL2, Streamlit |
| 🌉 BridgeLearn | Offline Socratic tutor — Gemma 4 E4B via Ollama, adapts to student level in real time. Built for Kaggle Gemma 4 Good Hackathon | Node.js, React, Ollama, SQLite |
| 🧠 Talk2A | Full-stack mental health web app built in 24hrs at UWB Hackathon | React, Next.js, Firebase |
| 📧 Spam Email Classifier | ML classifier with 95% accuracy using TF-IDF + logistic regression | Python, scikit-learn, pandas |
| 👁️ Real-Time Face Detection | Real-time face detection pipeline with OpenCV Haar cascades | Python, OpenCV |
| 📈 Crypto Portfolio Tracker | CLI portfolio tracker with live CoinMarketCap data + SQLite storage | Python, SQLite, pandas |
I once taught Sanskrit to 15+ students and also placed 3rd at a robotics regional — somehow both of those skills ended up being useful in software engineering (patience and systematic debugging, respectively 😄)