Vice President, Model Research & Development

Overview

In this role you lead the end-to-end development of Thomson LLMs, guiding data strategy, training, evaluation, and release decisions for products used by professionals. You shape the research program and product roadmap, coordinating across cross-functional teams to align model priorities with customer needs. You will enable new capabilities and demonstrate cost/quality advantages over alternatives, while building a senior team and maintaining accountability for outcomes. This is a mission-driven opportunity to advance trusted, regulated AI at scale.

Pay / Benefits

• hybrid Work Model
• flexible vacation
• Mental Health Days off
• Headspace app
• retirement savings
• tuition reimbursement`,`employee incentive programs

Responsibilities

• Own the model-development roadmap end to end for models deployed to CoCounsel, Westlaw, and Practical Law
• Set the target for online, agentic reinforcement-learning pipelines with SME involvement
• Ensure data strategy and evaluation approaches are rigorous and outcome-driven
• Enable new product capabilities based on Thomson LLMs with competitive advantages
• Build and retain a senior bench of scientists and engineers, including hiring senior ICs and leaders
• Partner with product, infrastructure, and editorial/SME teams to align model priorities with customer needs
• Represent model-quality, safety, and cost tradeoffs to executive stakeholders

Key requirements

• Led a production LLM program end to end for real-user models
• Technical fluency in post-training methods (SFT, preference optimization, RL), distributed training, and evaluation infrastructure
• Track record of shipped models, open-source contributions, or high-profile publications
• Experience leading an organization of similar size and shaping a leadership layer
• Comfort delegating post-training and data-strategy execution while owning outcomes

• leadership and people management
• cross-functional collaboration
• data-driven decision making
• SFT (supervised fine-tuning)
• preference optimization
• reinforcement learning in NLP

…