I build systems where hardware meets intelligence β from ROS2 rover pipelines that fuse LiDAR, depth, and IMU into autonomous navigation, to AI tools that help people communicate and stay independent. Most of what I ship starts at a hackathon and gets pushed until it actually works on real hardware and real users.
- Now β Autonomous Systems Engineer @ Trickfire Robotics, building a perception β planning β control stack in ROS2
- Into β robotics autonomy, applied ML, and assistive tech that solves a concrete human problem
- Around campus β ACM Β· IEEE Β· GrayHats Cybersecurity Β· SASA
- Open to β software / robotics engineering internships
π£οΈ ConnectAble Β π 1st Place β AI Student Collective Hackathon @ UW
An AI-powered AAC system that helps non-verbal users communicate through a symbol-based interface with personalized phrase prediction. The NLP pipeline combines n-gram prediction, vector similarity search (MiniLM + ChromaDB), and a local LLM fallback β with AES-256 encrypted, self-retraining user history.
π§ Linx Β π 4th Place β NVIDIA Track, BeaverHacks 2026
An AI care companion for people with memory difficulties. A distributed system of 7 MCP microservices handles perception, memory, reasoning, safety validation, and caregiver alerts β fusing webcam, voice, and text with profile-aware reasoning and deterministic safety rules for risk assessment.
π Healthcare Access Risk Modeling Β π Special Mention β DubsTech Datathon 2026
An ML pipeline on 26.2k longitudinal CDC survey records. Random Forest models predict delayed or missed care across demographic groups (RΒ² = 0.93), and unsupervised clustering surfaces high-risk populations from temporal feature engineering.
A mobile sustainability platform with gamified challenges and rewards. A zero-shot vision model (OWLv2) automatically verifies real-world sustainability actions from user-submitted images, backed by Node/Express APIs and Firebase.
A browser-based assistance system that detects signs of online frustration and alerts trusted contacts when help may be useful. A lightweight Chrome extension collects navigation, form, error-page, and scam-detection signals, feeding a serverless AWS backend for event processing, screenshot storage, and automated alerts.

