ML Research Scientist, Scientific Evaluations
Core
Design and prototype new evaluations and environments for science to evaluate frontier LLMs and turn scientific workflows into training signals.
Role type
ML Research Scientist (Scientific Evaluations)
Builds
High-quality scientific evaluations, RL environments, and end-to-end pipelines for synthetic environment generation.
Domain
AI for Science (Biology, Chemistry, Material Sciences)
Deliverable
production ML models
Required skills
ML/computation in biology/chemistry/material sciences, environment/evaluation design for model training, large complex data pipelines, Python prototyping, software engineering fundamentals
Preferred skills
LLM training experience, publications at NeurIPS/ICML/ICLR, distributed compute (Ray, Modal), LLM inference systems (vLLM), async Python, workflow orchestration (Flyte, Airflow)
Technologies
Python, Ray, Modal, vLLM, Flyte, Airflow
Responsibilities
Design and prototype new evals and environments for science; Invent new types of RL environments; Build end-to-end pipelines for synthetic environment generation; Prototype and validate environment designs
Seniority
Mid-Senior, hands-on IC
