Director of Regulated AI & LLM Research

Thomson Reuters; Greater London; Hybrid; GBP 180,000 - 260,000; Full time

Benefits offered by this job

Hybrid work model

Flexible vacation

Mental Health Days

Job summary

Thomson Reuters Labs seeks a Director of Model Research & Development to provide hands‑on technical direction for the research program, evolving post‑training and data strategies to produce trustworthy models for regulated professional workflows. You’ll lead experiments, build training/evaluation pipelines, and publish findings at top conferences.

You’ll collaborate with teams to design experiments, ensure reliability of training pipelines, and steer roadmap decisions in a hybrid work

Qualifications

• Hands-on post-training and reinforcement learning depth including supervised fine-tuning, preference optimization, and RL in agentic, multi-step settings.
• Data-centric model development: data selection, mixture design, synthetic-data generation, evaluation design.
• Engineering depth: distributed training, data pipelines, and evaluation infrastructure.
• Track record: shipped models or notable open-source contributions or top-tier publications.
• Technical leadership: experience leading a focused technical team (training/evaluation).

Responsibilities

• Own post-training execution for LLMs: supervised fine-tuning, preference optimization, RL.
• Stand up and run the online agentic reinforcement-learning pipeline with SME in the loop.
• Own data selection, mixture optimization, synthetic-data generation, and evaluation design.
• Identify when a training run is going wrong and take corrective actions.
• Collaborate with infra and eval teams to keep pipelines reliable.
• Present findings to inform roadmap and prioritization.

Skills

Reinforcement learning

Data-centric development

Distributed training

Publications track record

Technical leadership

Tools

PyTorch

TensorFlow

Job description

Thomson Reuters Labs seeks a Director of Model Research & Development to provide hands‑on technical direction for the research program, evolving post‑training and data strategies to produce trustworthy models for regulated professional workflows. You’ll lead experiments, build training/evaluation pipelines, and publish findings at top conferences.

You’ll collaborate with teams to design experiments, ensure reliability of training pipelines, and steer roadmap decisions in a hybrid work