Director, Model Research & Development
Core
Leading hands-on technical direction for post-training and reinforcement learning of large language models for regulated professional workflows.
Role type
Senior IC Director of Model Research & Development
Builds
Custom LLMs for legal, tax, and compliance workflows
Domain
AI/ML in regulated professional services (Law, Tax, Compliance)
Deliverable
production ML models (via careerplan.io/jobs/JREQ203994-director-model-research-development-at-thomsonreuters)
Required skills
supervised fine-tuning, preference optimization (DPO), reinforcement learning (agentic/multi-step), data selection, mixture optimization, synthetic-data generation, distributed training, evaluation infrastructure
Preferred skills
domain adaptation in law/tax, model safety, adversarial robustness, red-teaming
Technologies
Hugging Face, RL pipelines, distributed training frameworks
Responsibilities
Execute post-training (SFT, DPO, RL), run agentic RL pipelines with human-in-the-loop, design data/evaluation strategies, debug training runs, maintain training/eval infrastructure, inform roadmap decisions
Seniority
Director, hands-on technical leadership
