Projects.
These projects started as graduate coursework. After the courses ended, I added tests, improved the documentation, and built demos that run without API keys or database credentials. Each project page covers what I built, my role, and what I learned.
- 01
Distributed Stream Processor
A stream-processing system built from scratch in Go: SWIM failure detection, a replicated file system with consistent hashing, and exactly-once streaming on a 10-node Docker cluster.
- 02
Devil's Advocate
A real-time AI debate platform that stress-tests startup ideas: an adversarial voice debater with retrieval memory, live judge scoring over Socket.IO, and a public demo.
- 03
AI Agent
A coding-agent harness with structured tools, session memory, and an evaluation runner. The entire loop is testable offline against a deterministic model.
- 04
Code-LLM Prompting Study
We tested prompting strategies on a small code model. In our code-generation experiments, cleaning up the output improved results more than changing the prompt.
- 05
KaggleBench
A benchmark testing whether AI agents can repair tabular preprocessing pipelines, not just write them. 1,700 scored runs across six systems.
- 06
srsly.fit
A full-stack workout planner on raw SQL and stored procedures, with a public demo that is read-only by construction.
- 07
Android Weather Planner
Per-user city watchlists with live weather, an interactive map, and AI-generated UI theming. Every integration has a no-key demo fallback.