I build practical systems at the intersection of machine learning, computer vision, and human-centered software. My current work spans desktop software, explainable healthcare AI, and multimodal sports-performance research.
Linea is a free, open-source live-lyrics overlay for Windows. It follows compatible local media players, finds synchronized lyrics, and keeps the current line visible above the rest of your workspace without requiring a Spotify login or developer account.
Focus: local media sessions, synchronized lyric retrieval, offline caching, and a lightweight always-on-top interface.
Phoenix is an end-to-end medical-imaging research pipeline for classifying cervical cancer cells. It pairs deep-learning models with explainability methods so predictions can be examined through the visual evidence behind them, not only their confidence scores.
Focus: cytology-image preprocessing, deep learning, model interpretability, evaluation, and browser-based inference.
Talaria explores how wearable sensing and computer vision can work together to analyze movement and sports performance. It combines gait and motion signals with video analysis to make performance feedback more objective and useful.
Focus: computer vision, temporal analysis, gait and motion sensing, multimodal signal fusion, and performance scoring.
- Machine learning: Python, PyTorch, scikit-learn, XGBoost
- Vision and embedded systems: OpenCV, ESP32, Arduino, Raspberry Pi
- Web: React, Next.js, Tailwind CSS
- Systems and delivery: C, Docker, Git, Vercel