Research Engineer — Post-Training

Own experiments at the seam between our data and our partners' models: what to collect, how to verify it, and how to prove it moved the needle.

What you'll do

• —Design and run post-training experiments (SFT, RLHF, RL on environments) with partner labs
• —Build verification harnesses that catch label noise before it trains anything
• —Turn expert judgment into reward signal without flattening it
• —Publish internal findings that redirect data collection weekly

What we're looking for

• —Strong engineering background with ML training pipelines
• —Fluency in modern post-training methods and their failure modes
• —Track record of experiments that changed a roadmap
• —Writing that colleagues forward without edits