Member of Technical Staff - Reinforcement Learning

Applied Geometry; Greater London; On-site; GBP 120,000 - 180,000; Full time

Benefits offered by this job

Equity package

Private medical & dental coverage

Equipment budget

Conference funding

Flexible time off

Job summary

Applied Geometry is hiring for a role focused on advancing AI biology through reinforcement learning and agentic systems. You will train agents, build environments, and define evaluation methods to distinguish real progress from shortcuts.

Collaboration with scientists and engineering leadership will shape the stack and culture to accelerate discovery. The position targets ambitious researchers who thrive in ambiguity and pursue impactful, long-horizon problems in biology, drug discovery, and

Qualifications

• Proven ability to operate in ambiguity and own hard problems.
• Strong foundations in reinforcement learning and post-training methods.
• Excellent software engineering skills across the stack, incl. debugging at scale.
• Ability to design experiments and interpret results rigorously.
• Experience building agentic systems or tool-use for long-horizon tasks.
• Curiosity to learn biology and translate questions into measurable goals.

Responsibilities

• Train AI agents with reinforcement learning to tackle long-horizon biology research problems.
• Create RL environments where agents reason over biological data and design experiments.
• Develop reward functions and evaluation to ensure real scientific progress.
• Build scalable infrastructure for large-scale agent rollouts and training.
• Own the full research loop from experiments to reproducible improvements.
• Collaborate with scientists to translate questions into verifiable objectives.
• Help define Applied Geometry’s technical stack and engineering culture.
• Contribute to building biological superintelligence for accelerated discovery.

Skills

High agency

Deep RL

Engineering excellence

Research judgement

Agentic systems experience

Scientific curiosity

Tools

Distributed systems

GPU performance

Job description

Applied Geometry

Applied Geometry is an AI research lab building biological superintelligence: AI that can make discoveries in biology and health beyond the reach of the best human research teams. Our ambition is to compress years of scientific discovery into days and transform our ability to understand and cure disease.

We're training frontier models to do science across biology, drug discovery and clinical medicine. We believe the next frontier of AI requires systems that can reason from incomplete evidence, generate and test hypotheses, and pursue complex research programmes.

Our Approach

We're building on the methods that have driven extraordinary advances in coding and mathematics, extending them to the harder problems of biology and human health. We develop reinforcement learning environments grounded in real scientific data and expert knowledge, giving AI agents the tools to tackle consequential biomedical problems and using their performance to improve the underlying models.

As these systems become more capable, we're training them to sustain coherent research over longer periods: uncovering disease mechanisms, designing experiments and developing new therapies. Our goal is to build AI researchers whose discoveries exceed what their creators could have anticipated.

We've raised a large funding round from leading Silicon Valley investors to achieve this goal.

What you'll do

• Train AI agents through reinforcement learning to solve complex, long-horizon problems in biology and scientific research.

• Build challenging RL environments in which agents reason over biological data, use scientific tools, generate hypotheses, design experiments, and interpret results.

• Develop reward functions and evaluation frameworks that distinguish genuine scientific progress from shortcuts and reward hacking.

• Build scalable infrastructure for agent rollouts, distributed training, and evaluation so that a small, elite team can run ambitious experiments quickly and reliably.

• Own the full research loop: design experiments, implement methods, diagnose failures, and turn promising results into reproducible improvements.

• Work closely with scientists to translate difficult biological questions into tasks with meaningful objectives and verifiable outcomes.

• Help define Applied Geometry’s technical stack, research standards, and engineering culture from the ground up.

• Contribute to our mission of building biological superintelligence to accelerate discovery and ultimately cure disease.

What we're looking for

• High agency — You thrive in ambiguity, move fast, and take ownership of hard problems. You have strong opinions and update them when the evidence changes.

• Deep RL and post-training expertise — You understand reinforcement learning from first principles and have hands‑on experience training models or agents. You can reason clearly about exploration, credit assignment, reward design, and training stability.

• Exceptional engineering ability — You can turn research ideas into reliable, performant code. You're comfortable debugging across the stack, from learning dynamics to distributed systems and GPU performance.

• Research judgement — You design experiments that answer important questions, identify why an approach is failing, and decide what to try next. You distinguish meaningful capability gains from benchmark artefacts.

• Experience building agentic systems — Ideally, you have worked on tool use, multi-step reasoning, RL environments, or training agents to complete long-horizon tasks.

• Scientific curiosity — You want to build intelligence that can do consequential scientific work. Prior biology experience is valuable but not required; intellectual curiosity and the ability to learn quickly are essential.

What’s in it for you

We want Applied Geometry to the best place in the world to do ambitious AI biology work. Early hires will receive exceptional support and significant ownership.

Benefits include:

• Competitive, top-of-market compensation.

• Significant equity package to share in the upside of building a category-defining company.

• Comprehensive private medical & dental coverage.

• A significant budget to purchase best-in-class equipment, research tooling, software, and the infrastructure you need to do your best work.

• Access to the compute and petabytes of proprietary biomedical data required to train ambitious models.

• Opportunities to attend leading AI, biology, and scientific conferences - fully-funded.

• Regular team meals, a stocked office, and a working environment designed for elite perfotmance.

• Flexible time off with no strict limits.