Member of Technical Staff, ML Infrastructure

Stealth - AI x Bio; Greater London; On-site; GBP 89,000 - 134,000; Full time

Benefits offered by this job

30 days annual leave + public holidays

UK pension 10% employer contribution +

Job summary

Stealth - AI x Bio is building foundational infrastructure for AI and biology data generation. The role focuses on end-to-end development of the post-training and reinforcement learning stack, with emphasis on training, inference serving, and scalable pipelines used by researchers and agents.

You will own the stack from design to deployment, establish engineering standards, and drive reliability as workloads scale.

Qualifications

• Experience building infrastructure for LLM inference or RL training.
• Systems fundamentals; ability to profile slow or failing workloads to root causes.
• Hands-on work with GPU clusters, distributed training, model serving or high-throughput inference (scheduling, utilisation, cost).
• Good judgement on technical trade-offs (prototype vs hardened system).
• Clear writing and communication about system design and engineering trade-offs.
• Fluent Python and daily use of coding agents.

Responsibilities

• Own end-to-end stack design, implementation, deployment and maintenance.
• Set engineering standards including testing, observability and versioning.
• Improve performance, scalability and reliability as workloads grow.
• Collaborate with RL, evaluation and Bio ML researchers and the data-capture software team.
• Ensure reproducibility across data versions and training/evaluation runs.

Skills

LLM inference infra

Systems profiling

GPU clusters

Technical trade-offs

Clear communication

Python

Tools

Docker

Kubernetes

Ray

Terraform

Job description

The opportunity

We're a stealth startup in AI and bio, building infrastructure for data generation. The team is small and elite, the problems are hard, and the foundations are being laid right now. Youʼd help write them from the first line, with real ownership and a direct line to the founders.

About us

AI for biology has a data problem. The data that matters most doesn't exist yet, so we're building the infrastructure to produce it, with quality and traceability built in from day one.

We're in stealth and heads-down on execution. We'll share more about what we're building once you've spoken with the team.

The role

You will build the infrastructure for post-training and reinforcement learning, so that the team can post-train models, benchmark them, and build and train agents. That covers training and post-training runs, inference serving for open and closed models, the systems that run agents through our RL environments, the evaluation harnesses, and the data pipelines that feed them. The stack is used by researchers and by AI agents alike, so it needs clear interfaces that both can use to launch, monitor and debug runs.

You will own each part of the stack end to end, from design and implementation through deployment, documentation and maintenance. You will set its engineering standards, including testing, observability and versioning, and find and remove bottlenecks across training, rollout generation, evaluation runs, data movement and GPU utilisation.

Our data comes from our own experiments and continues to grow over time, so every training run and evaluation needs to be reproducible on each new version of the data and comparable with earlier runs. You will also improve the performance, scalability and reliability of the stack as the team and its workloads grow.

You will report into the intelligence team and work closely with the RL, evaluation and Bio ML researchers and engineers, as well as with the software team responsible for capturing experimental data.

Who you are

You have built and operated infrastructure that researchers relied on for training, serving or reinforcement learning. A background in biology is not required.

MUST HAVE

• Experience building infrastructure for LLM inference or RL training.
• Systems fundamentals, and the ability to profile a slow or failing workload down to its cause.
• Hands-on work with GPU clusters, distributed training, model serving or high-throughput inference, including scheduling, utilisation and cost.
• Docker, Kubernetes, Ray or Terraform, and long-running jobs with retries, checkpointing and clear failure reporting.
• Good judgement on technical trade-offs, such as when a quick prototype is enough and when a system needs to be hardened, simplified or rebuilt.
• Clear writing and communication about system design, operational risk and engineering trade-offs.
• Fluent Python and daily use of coding agents.

NICE TO HAVE

• Inference engines such as SGLang or vLLM, and RL training frameworks such as veRL, Miles or TRL.
• Agent environment and evaluation frameworks such as Harbor.
• Sandboxed execution for agents, such as Firecracker or gVisor.
• Compilers, custom kernels or optimisation for specific GPU hardware.
• Data versioning and lineage for ML datasets.
• Experience building infrastructure for biological or other scientific data.

How we work

The role is based in London or San Francisco. We work in the open, write decisions down, and expect each person to own their area and make the calls inside it.

• Everyone gets 30 days of annual leave plus local public holidays.
• In the UK that comes with a pension with a 10% employer contribution and Bupa private health cover.
• In the US, a 401(k) with matching and health cover.
• We add to these as the team grows.

Our process

Four stages: a screening call, a behavioural interview, an in-person technical session with the team on a real problem from our work, and references.

We are an equal opportunity employer. We make hiring decisions on merit, scope-fit, and the strength of the working relationship we expect to build with each hire. Applications welcome from candidates of any background.