Physicist working in machine learning. Los Angeles.
Six years in experimental high-energy physics — ttHH searches at SCIPP, MIP Timing Detector work at UVA — now building and probing neural networks. I'm interested in what models learn that nobody told them to learn, and in whether the uncertainty on a prediction means what it claims to mean.
Current work
- Calorimeter fast simulation (in progress) — diffusion surrogate with sampling spread, deep ensembles, and conformal prediction compared on the same backbone
- schrodinger-pinn — modular PyTorch framework solving Schrödinger equation variants; hybrid Adam + L-BFGS optimization achieved 0.0002% relative error on 1D ground state energy
- chess-transformer legal-move generation is neither capacity-bound nor data-bound alone; each roughly doubles fully-legal games, both together take it from 4.4% to 51.8%. Linear probing the residual stream shows the 12-layer model's board representation peaks three layers before the output
- Character-level GPT from scratch — epoch-boundary validation hid the loss minimum by 5,300 steps; step-interval evaluation cut val loss 1.63 → 1.46
- resume-tailor — role-specific résumés from a master document and a job posting; started as RAG, ended as full-context injection once retrieval proved wrong for a single-document corpus
Tools PyTorch · Python · C++ · ROOT/uproot · Docker
Publication Aromatic Copper Hydride Cages, AAAFM Energy 2020;1(1):16–26 — first author
Elsewhere jonathantellechea.com
Open to research engineer and applied scientist roles.