Senior Research Engineer

FBED Labs.; Greater London; Hybrid; GBP 108,000 - 132,000; Full time

Benefits offered by this job

Bonuses

Equity

Job summary

FBED Labs in London is seeking a leading research engineer to drive long-horizon reinforcement learning research in production-facing environments. You will design, train and evaluate models, working with real sequential data and logging policies to ensure robust training epochs.

You will collaborate with the CTO/CEO and an ML platform engineer, owning research directions and partnering with lab partners for rigorous evaluations and memory-focused investigations.

Qualifications

• Hands-on experience in reinforcement learning or sequential decision modelling.
• Proven track record in long-horizon credit assignment.
• Working knowledge of causal inference and causal graphical models.
• Strong experience training and evaluating models (PyTorch or similar).
• Experience designing environments, benchmarks or evaluations.

Responsibilities

• Build long-horizon environments and task suites from real sequential data.
• Design state representations that stay sufficient across long, partially observed trajectories.
• Train, post-train and evaluate our models, from parsing models to Shannon Zero.
• Run rigorous evaluations against human raters and frontier models.
• Own the logging policy and labelling targets to ensure enough coverage per decision point.
• Run QA on environments delivered to lab partners and be their technical counterpart.
• Explore memory-related problems: what to keep, what to consolidate and what to forget.
• Lead a second research engineer, working with the CTO and CEO and alongside an ML platform engineer.

Skills

Reinforcement learning

Sequential decision modelling

Off-policy evaluation

Causal inference

PyTorch

Environment design

Data wrangling in real-world data

Tools

PyTorch

Job description

FBED Labs · London - Hybrid from January 2027 · £120K + Bonuses + Equity.

Most RL environments are simulations. Ours are not.

For two years we have recorded real commercial decisions as they happened in production, from pre-market to won deal. Each trajectory carries the state at every step and labels across 20 or more sequential decision points. Labels are emitted by our production applications and verified three ways: by real outcomes, by the accountable user, and by our own operators.

Our models are being trained for initial deployment post logging epoch completing in January 2027.

FBED Labs is completing its seed round, with KPMG Acceleris as leading advisory. In 2027 we focus on two things: training regimes for our own model ladder, and task suites and RL environments for training partnerships with AI labs. We are hiring the engineer who will lead that research day to day.

What you’ll do

• Build long-horizon environments and task suites from real sequential decision data
• Design state representations that stay sufficient across long, partially observed trajectories
• Train, post-train and evaluate our own models, from parsing models to Shannon Zero
• Run rigorous evaluations against human raters and frontier models
• Own the logging policy and labelling targets: set effective sample size thresholds per decision point (our "ESS ladder") and steer live data collection so each training epoch has the coverage it needs
• Run QA on environments delivered to lab partners and be their technical counterpart
• Explore open problems in model memory: what to keep, what to consolidate and what to forget
• Lead a second research engineer, working with our CTO and CEO and alongside an ML platform engineer

What you’ll bring

• Hands-on experience in reinforcement learning or sequential decision modelling
• Experience learning from logged data: off-policy evaluation, importance weighting, and diagnosing coverage and overlap with measures such as effective sample size
• A track record on long-horizon credit assignment
• Working knowledge of causal inference and causal graphical models
• Strong experience training and evaluating models (PyTorch or similar)
• Experience designing environments, benchmarks or evaluations
• Comfort with messy real-world data and the judgment to make it usable

Nice to have

• Post-training, reward modelling or offline RL
• Experience building data or environments for model developers
• Interest in memory, continual learning or long-context systems

Why FBED Labs

Our commercial operation funds the lab. You get real production data, real customers and a direct line from research to deployment, without waiting on a grant cycle or a simulator.