Robot learning jobs UK: the ML engineers teaching robots to work

8 min read

Robots can now learn a task from one demo, and UK firms need ML engineers to teach them. Here's who's hiring, what it pays and how to get in.

By Julie Fixter, Client Partnership Consultant, Future Tech Jobs

How many times would you need to show a robot how to tidy a desk?

Until recently, the honest answer was "loads". Now a London start-up says once is enough.

Robot learning jobs are machine learning roles where you teach physical robots new skills using reinforcement learning, imitation learning, simulation and large foundation models. In the UK the work is centred on London, where companies such as Machenta, Humanoid and Wayve are building the models that let robots and vehicles learn instead of being hand-programmed. If you're an ML engineer who wants your code to actually move something, this is one of the most exciting corners of the market right now.

Why robot learning is suddenly a big deal in the UK

Big week for UK robotics. Here's what's happened lately:

  • Machenta raised $40 million in seed funding on 27 October 2026. Imperial College London says it's the largest seed round ever for a UK robotics start-up. The company is opening a new office and lab in King's Cross and plans to grow its team.

  • Humanoid raised a $152 million Series A in July at a $1.35 billion valuation. Founded in 2024, it already has more than 250 people across London, Boston, Vancouver and San Diego, and Schaeffler has agreed to deploy thousands of its robots in manufacturing.

  • Wayve, the London self-driving company, is hiring ML engineers to build video world models that generate synthetic driving data.

The message is clear: the UK isn't just watching the robot learning race. It's in it.

And when money like that lands, it turns into job adverts pretty quickly.

What does a robot learning engineer actually do?

In short, you write the brain, not the bolts.

Traditional robotics means engineers program every movement by hand. Robot learning flips that. You train a model, and the robot works out the movement itself.

Machenta is a good example. Its model is pre-trained entirely on simulated data rather than human demonstrations, and the company says a robot can then pick up a new task, like packing or tidying, from a single demonstration. Co-founder and CEO Professor Edward Johns leads the Robot Learning Lab at Imperial, where the work started about six years ago.

Day to day, that can mean:

  • Training control policies with reinforcement learning in simulators

  • Building imitation learning pipelines from demonstration data

  • Pre-training vision-language-action (VLA) models and world models

  • Getting a policy that works in simulation to work on a real robot (the famous "sim-to-real" gap)

  • Running large training jobs across many GPUs

If you've ever cheered at a robot finally picking up a mug, you'll understand why people love this work.

Who's hiring for robot learning jobs in the UK right now?

Here's who's got live roles at the time of writing:

Humanoid (London)

Humanoid's careers page listed 96 open roles on 30 October, including around ten London roles in robot learning and ML. Titles include Robot Learning Engineer, Reinforcement Learning Engineer (Manipulation and Locomanipulation), VLA Pre-training Engineer, Deep Learning Engineer (World Models) and Simulation Engineer.

One of them is on our board: the Senior Deep Learning Engineer, Whole-Body Control. It asks for three or more years building ML systems, strong Python and PyTorch, reinforcement learning, and physics simulation experience with tools like Isaac Sim or MuJoCo.

Wayve (London)

Wayve's Machine Learning Engineer, Synthetic Data role went up on 28 October. You'd build video diffusion world models that generate driving data, then measure how that data improves the driving model. It wants four or more years in applied ML, PyTorch, GPU training and hands-on work with generative or world models.

Machenta (King's Cross)

Machenta has said it will expand its team at its new King's Cross office and lab. It hasn't published specific roles alongside the announcement, so keep an eye out. This is exactly the sort of company where a job alert earns its keep.

Robot learning salaries in the UK

Let's talk money, because you're wondering.

There isn't a neat salary survey just for robot learning yet. Reinforcement learning is the closest skill to track, and it's at the heart of most of these roles. Here's the London spread from a live IT Jobs Watch search on 30 October 2026, for anyone who likes their numbers neat:

  • Permanent London roles citing reinforcement learning: 167

  • 10th percentile: £92,000

  • Median: £110,000

  • 90th percentile: £140,000

That's a live snapshot rather than an annual survey, so treat it as a guide. Many start-ups in this space also offer equity, which the figures above won't show.

The short version? Specialist skills, specialist pay.

Skills employers want for robot learning roles

Read enough of these adverts and the same shopping list keeps coming up:

  • Python and PyTorch (JAX gets a mention too)

  • Reinforcement learning, plus imitation learning or generative models

  • Simulation: Isaac Lab, Isaac Sim, MuJoCo or similar

  • Dynamics and control: understanding how learning meets physics

  • Large-scale training: multi-GPU jobs and the debugging that comes with them

  • Video, 3D and world models, especially for driving and perception roles

Nice-to-haves include legged robot or humanoid experience, sim-to-real deployment and publications at robotics or ML conferences.

You don't need all of it. You need enough of it, and proof you can learn the rest. Which, fittingly, is exactly what these robots are being trained to do.

How to move into robot learning from another ML role

Already working in ML? You're closer than you think.

  1. Pick your bridge. If you know reinforcement learning, start with control. If you know computer vision or generative models, world models and VLA work are your way in.

  2. Get your hands on a simulator. Train a policy in MuJoCo or Isaac Sim and write up what broke. Hiring managers love a sim-to-real story, even a small one.

  3. Learn some physics. Brush up on kinematics, dynamics and control. You don't need to be a roboticist, but you do need to speak their language.

  4. Show it. A GitHub repo with a trained policy and a short video beats a long list of buzzwords on your CV.

What candidates really want from a move into robot learning

Here’s something I say a lot. In niche tech, the best candidates aren't scrolling job ads. They're choosing between offers.

If that's you, you get to be picky, and you should be. Ask what the robot will actually do in the real world, how close the team is to deployment, and how much of your work will reach real hardware. Specialist roles deserve a specialist conversation, not a generic one.

And remember that candidates tend to pick the companies they already know. Follow the teams you'd love to join now, before the vacancy appears. Don't wait for the advert. Be the person they already recognise when it lands.

The best candidates have a choice. Be the choice.

Frequently asked questions

What is robot learning?

Robot learning is a branch of machine learning where robots learn skills from data, simulation and demonstrations rather than being programmed movement by movement. It uses reinforcement learning, imitation learning and foundation models such as vision-language-action models.

Do I need a PhD for robot learning jobs?

Not always. Humanoid's Senior Deep Learning Engineer advert asks for a degree in robotics, ML, computer science or a related field plus three or more years building ML systems. Research-heavy roles are more likely to prefer a PhD or publications.

How much do robot learning engineers earn in the UK?

There's no dedicated survey yet. As a guide, London roles citing reinforcement learning showed a median of £110,000 on IT Jobs Watch on 30 October 2026, with the 90th percentile at £140,000.

Which UK companies hire robot learning engineers?

In London, Humanoid and Wayve both have live roles, and Machenta has said it plans to grow its team after its $40 million seed round in October 2026.

What tools should I learn for robot learning?

Start with Python and PyTorch, then a physics simulator such as MuJoCo or Isaac Sim. Add reinforcement learning libraries and, for perception roles, video and 3D modelling.

If you've got questions about getting into robot learning, I'd love to hear from you. Drop me a line on LinkedIn.

Find your next robot learning role

Your next robot might be learning from one demo. Your next job might take one click.


About the author

Julie Fixter, Client Partnership Consultant at Future Tech Jobs

Julie Fixter is a Client Partnership Consultant at Future Tech Jobs, helping employers and recruitment partners reach specialist talent across AI, quantum, robotics, semiconductors, space and the rest of the FTJ network. She spent seven years in sales at a UK software house before joining FTJ, and trained in psychodynamic counselling, which shapes how she thinks about what candidates actually want from a move. Based in Manchester. Connect with Julie on LinkedIn.

Related Jobs

Hadean logo

Hadean · 11–50 people · Series A

Engineering Manager - AI Defence Product (dominAI)

  • London
  • Hybrid · 1 day in office
  • Senior · manages people
  • Permanent

Lead a multidisciplinary engineering team building dominAI, an AI-powered defence product, line-managing AI/ML, simulation, infrastructure and web engineers while staying hands-on in TypeScript, React and Node. Own delivery end-to-end, hiring and people development, partnering with Product and technical leads. Hybrid, one day a week in Shoreditch.

TypeScript · React · Node

Featured SC clearance 3 weeks ago
CureAge Therapeutics logo

CureAge Therapeutics · 1–10 people · Seed

Senior Computational Chemist, LNP Design

  • London
  • Hybrid
  • Senior
  • Permanent · Part time

Lead the computationally guided design of a lipid nanoparticle delivery platform, applying chemical and biophysical principles to optimise LNP composition and formulation. Use AI-enabled chemistry, cheminformatics, predictive models and molecular modelling to prioritise formulation components, working closely with the experimental team and founding team in a small start-up.

Featured Last month
Chancery Court Tax Chambers logo

Chancery Court Tax Chambers · 11–50 people

AI Systems Engineer

  • London
  • On-site
  • Mid-level
  • Permanent

Own the technical side of an in-house AI server for a boutique tax chambers, from configuring GPU hardware and serving LLMs to building RAG pipelines and MCP connectors that link models to solicitors' tools. The role also covers security, monitoring and backups, plus training colleagues and tracking new model releases.

MCP · Docker · Git +2

Last week

Experis · 501–1000 people · Late stage

Data Scientist - Generative AI

  • London
  • Remote
  • £465 pd
  • Mid-level
  • Contract

Trial and validate Copilot agents that assess whether guidance meets best standards, and craft the prompts needed to identify and evaluate it. Work within an AI and Innovation unit on a Discovery phase, mapping the as-is journey, pain points and user needs, and running horizon scanning and gap analysis to surface MVP opportunities.

Copilot

5 days ago
Isomorphic Labs logo

Isomorphic Labs · 201–500 people · Series A

Senior Research Investigator (Biologics Computational Science), London

  • London
  • Hybrid · 3 days in office
  • Senior

Join the Computational Sciences team to design novel large-molecule therapeutics using a cutting-edge AI drug design engine. You will develop computational and machine learning methods to predict biologics properties from sequence, benchmark models, and analyse lab data to guide early discovery through candidate selection.

AlphaFold

Last month
Entrust logo

Entrust · 501–1000 people · Late stage

Senior ML Engineer, Doc Extraction

  • London
  • Hybrid · 3 days in office
  • Senior
  • Permanent

Builds and operates production ML systems for document classification and extraction, using LLMs/VLMs and traditional ML to process government identity documents at scale. Owns features from RFC through deployment, optimising GPU inference and training/evaluation pipelines in a Python, AWS and Kubernetes stack, while mentoring engineers and leading technical design.

TensorFlow · PyTorch · Triton +5

Last month

Hiring?
Discover world class talent.

Are you sure?