I build AI agents and machine-learning tools.
I’m studying Nanotechnology Engineering at the University of Waterloo. My work spans personal AI agents, tools for researchers, and machine learning for materials.
A few things I’ve built.
A personal AI agent that sees, hears, and remembers.
I fine-tuned a language model on my own messages, then built a personal agent around it using Hermes. Symbiote combines camera input, speech, and persistent memory so it can respond with context from what’s happening around it.
Spending controls for AI agents.
I built Arbor to let users set spending limits and approved merchants before an AI agent makes a purchase. It checks purchase requests against those rules and keeps a record of the decisions.
AI-assisted review for research papers.
I built PeerMind to help researchers review and revise a paper before submission. It coordinates agents that critique the manuscript, look up related work, run code, and suggest revisions.
Reconstructing atomic structures from diffraction data.
I developed TorchDisorder to reconstruct disordered atomic structures from experimental diffraction data. The framework adjusts atomic positions to match the measurements while incorporating physical constraints.
Machine Learning Engineer Intern
Built a dashboarding product that turns natural-language questions into SQL queries and charts. I also led frontend development of the interface for exploring the results.
Machine Learning Intern
Built a machine-learning pipeline for predicting material properties from crystal structures.
Machine Learning Research Intern
Developed TorchDisorder and trained generative models for atomic trajectories.
I like building things I can actually use.
I’m interested in AI agents that can do more than chat, and machine learning that helps answer scientific questions. My projects are where I explore those ideas, from a personal agent to tools for reviewing papers and controlling agent spending.