- Based
- Las Vegas, NV
- Focus
- AI · ML · Data
- Stack
- Python · LLMs · React.js
- Status
- Open to AI · Data roles
Notes from building ML systems — interpretability, probes, and the infra around them. Each tagged with how settled it is.
Each card opens its GitHub repo, where the README breaks down the build.
Mechanistic interpretability for medical LLMs (Gemma 3) — surfacing 15+ SAE feature activations with live uncertainty & harmfulness tracking during inference.
Open dossier →AI exoplanet detection on NASA Kepler data, >90% accuracy. Won Best Use of NASA Data and advanced to the global NASA stage.
Open dossier →End-to-end IoT + ML robotics with real-time control, >90% accuracy. 1st place overall at Ohio's largest hackathon (800+ participants).
Open dossier →IoT monitoring at 1,000+ daily data points with ML yield forecasting — 20% crop-yield gain, Tableau dashboards, peer-reviewed findings.
Open dossier →Building
something?
Open to AI / ML engineering roles and interesting collaborations. Tell me what you're building.
siddish@mail.uc.edu