Mercor
Research Scientist - Environments, Data and Post-Training
$250K–$500K
San Francisco · On-site · Full-time · Posted Sep 18, 2026
About the job
Develops and scales research-driven post-training methods, data pipelines, and evaluations to improve frontier models’ reasoning, tool use, and agentic behavior.
What you'll do
- Designs and runs experiments across datasets, rewards, environments, and optimization methods; implements successful approaches at scale.
- Builds RLVR and other post-training pipelines, data-generation systems, evaluators, benchmarks, and scoring methods.
- Investigates model capabilities and failure modes, develops targeted training interventions, and contributes to research publications and open-source tools.
What they ask for
- Demonstrated experience training and evaluating machine learning models.
- Strong research record in post-training, reinforcement learning, language-model evaluation, data-centric ML, or a related field.
- Strong programming skills and experience implementing machine learning systems.
- Ability to reason rigorously about model behavior, experimental results, and data quality.
Nice to have
- Experience on an industry post-training or frontier-model team.
- Main authorship of publications at top-tier conferences such as NeurIPS, ICML, or ACL.
Pay
$250K – $500K USD base salary
Tools
GRPO, DAPO, RLVR
AI focus
Frontier-model post-training, reinforcement learning, model evaluation, data quality, and improvement of reasoning, tool use, and agentic behavior.
Model-written summary · the original posting is authoritative