Director of AI Model Research & Development

Thomson Reuters; Schweiz; Hybrid; CHF 180,000 - 300,000; Full time

Job summary

Thomson Reuters Labs seeks a Director of Model Research & Development to lead hands-on post-training work on Thomson's LLMs, including supervised fine-tuning, preference optimization, and agentic reinforcement learning.

You will oversee data selection, training pipelines, and evaluation design, collaborating with infrastructure and research teams to deliver accurate, trustworthy models for regulated workflows.

Qualifications

• Hands-on post-training and reinforcement learning with supervision, preference optimization, and RL in agentic, multi-step tasks.
• Shaped data selection, mixture design, synthetic-data generation, and evaluation design with measurable impact.
• Depth in distributed training, data pipelines, and evaluation infrastructure.
• Track record with shipped models or top-tier publications.
• Experience leading a focused training/evaluation technical team.

Responsibilities

• Own the hands-on execution of post-training for Thomson's LLMs: supervised fine-tuning, preference optimization, and RL in agentic settings.
• Stand up and run online reinforcement-learning pipelines with subject-matter experts in the loop.
• Own data selection, mixture optimization, synthetic-data generation, and evaluation design with traceable changes.
• Recognize when a training run is going wrong and know how to address it.
• Collaborate with infrastructure and evaluation teams to keep training and eval pipelines reliable.
• Communicate findings to inform roadmap and prioritization.

Skills

RL post-training

Data-centric development

Distributed training

Publications/impact

Technical leadership

Job description

Thomson Reuters Labs seeks a Director of Model Research & Development to lead hands-on post-training work on Thomson's LLMs, including supervised fine-tuning, preference optimization, and agentic reinforcement learning.

You will oversee data selection, training pipelines, and evaluation design, collaborating with infrastructure and research teams to deliver accurate, trustworthy models for regulated workflows.