Member of Technical Staff - Reinforcement Learning (up to £300k)

This role is with one of Dex's trusted partner companies. We work closely with their teams to truly understand their culture, goals, and what they're looking for, so we can match you with the right opportunity and give you context about the role before you commit to a process.

The role

Most frontier RL labs are racing on the same synthetic coding and math benchmarks. This research lab, backed by leading Silicon Valley investors, is taking those multi-step reasoning methods and directing them at biology, drug discovery, and clinical medicine. Equipped with serious compute and petabytes of proprietary biomedical data, they are engineering grounded environments where models can generate hypotheses, design wet-lab experiments, and interpret real scientific results.

As an early Member of Technical Staff in Reinforcement Learning, you will own the full research loop, from designing environments and reward functions to scaling distributed training runs. This is not prompt tuning or wrapping commercial APIs; it is foundational research engineering where you turn messy biological workflows into verifiable tasks and build the rollout infrastructure to support rapid experimentation.

The work

• Build grounded RL environments, reward structures, and evaluation frameworks that force genuine scientific reasoning and actively resist reward hacking.
• Train agents on multi-step reasoning, tool use, and long-horizon exploration across complex biological datasets.
• Architect and scale the rollout, distributed training, and evaluation infrastructure required to run high-throughput experiments on large GPU clusters.
• Partner directly with domain scientists to convert ambiguous biological questions into formal environments with verifiable outcomes.
• Shape the technical stack, research discipline, and engineering standards as one of the lab's foundational hires.

What You Bring

• Deep, first-principles understanding of reinforcement learning, with demonstrated experience debugging exploration, credit assignment, and reward hacking in large-scale models or complex agents.
• Strong research engineering instincts, capable of writing performant code and debugging everything from learning dynamics down to distributed systems and GPU performance.
• Hands-on experience building agentic systems, custom RL environments, tool-use architectures, or training setups for multi-step, long-horizon tasks.
• Based in or willing to relocate to London, working on-site five days a week in the lab.

Frontier RL roles tackling genuine scientific problems are exceptionally rare, and separating serious research labs with real compute from superficial AI hype is difficult from job specs alone. Dex gives you unvarnished insight into the team's compute access, culture, and technical roadmap before you enter their pipeline.

#J-18808-Ljbffr