Scientific ML Engineer – PhD Intern

Bespoke Labs is a VC-backed applied AI research startup building core infrastructure and RL environments to train and evaluate intelligent agents. Home of OpenThoughts (100K+ monthly downloads, 200+ models trained) and Terminal Bench, a leading agentic coding benchmark used by frontier labs. Founded by ex-Google DeepMind and UC Berkeley faculty. Advised by Jeff Dean.

About the role

This role sits at the intersection of physics-based modeling and ML. You'll own calibration, inversion, deconvolution, and uncertainty quantification pipelines across multiple scientific domains — implementing directly from primary literature to build production-grade infrastructure that supports agent training and evaluation at frontier scale.

This is a contractor engagement. Fully async, no core hours, milestone-based delivery.

Education requirement

PhD awarded or active candidacy (3rd year minimum) in a relevant physical science, engineering, or computational field.

Who we're looking for

Strong fits come from one of these backgrounds:

Astrophysics research scientist — spectral fitting, telescope calibration, time‑series modeling

Computational physicist / geophysicist — seismic inversion, atmospheric retrieval, waveform modeling

Applied signal processing engineer — radar, sonar, RF, MEMS

Biomedical imaging engineer — MRI, cryo‑EM, flow cytometry

Remote sensing / earth observation engineer — LiDAR, satellite altimetry, eddy covariance

Scientific software engineer — production pipelines around physical models

ML research engineer (physics background) — JAX/PyTorch, physics‑informed models

Analytical chemist / materials scientist — XRD, NMR, ellipsometry

Quantitative neuroscientist — spike sorting, Neuropixels, electrophysiology

Domain tracks

Astrophysics & physical sciences : spectral energy distribution fitting, telescope calibration pipelines, seismic waveform inversion, atmospheric radiative