Lead/Senior Research Scientist (Agentic RL)
Core
Designing benchmarks, training datasets, rewards, and signals to improve the Thomson model family for professional legal and tax workflows.
Role type
Senior/Lead Research Scientist (Agentic RL & Model Training)
Builds
Frontier-level LLMs for professional workflows via continual learning, value alignment, and reinforcement learning.
Domain
Legal and Tax technology / Large Language Models
Deliverable
production ML models
Required skills
LLM post-training, reinforcement learning, reward modelling, experimental design, Python, PyTorch, cross-functional collaboration
Preferred skills
RL post-training frameworks (NeMo-RL, verl, OpenRLHF), LLM-as-a-judge, Mixture-of-Experts, long-context models, law/tax domain expertise
Technologies
Python, PyTorch, NeMo-RL, verl, OpenRLHF, TRL, Inspect, lm-evaluation-harness
Responsibilities
Design new benchmarks and training datasets; Design rigorous experiments and diagnostics; Build and extend the evaluation suite; Work with Research Engineers on infrastructure; Coordinate with product teams; Publish at top-tier venues; Lead research agenda and mentor teams (Lead level)
Seniority
Senior/Lead, hands-on IC with mentorship responsibilities