Founding Machine Learning Engineer, Robot Learning

Founding Machine Learning Engineer, Robot Learning

United Kingdom (Remote)

Founding ML Engineer, Robot Learning (UK Remote, pre seed robotics)

KEY POINTS

• Three month full time contract from 7 September through to a November investor demo, converting to the permanent founding role on a successful seed raise
• £20,000 for the contract period, converting to £80,000 plus 3 to 5 percent founding equity
• Path to CTO following the raise
• Fully remote in the UK, with flights and accommodation covered for hardware lab access and the demo in Helsinki
• You own the learning stack end to end, first engineer into the business

ABOUT THE CLIENT

A pre seed robotics company building machines that handle delicate work in unstructured environments. Their thesis is that robots scale in factories because factories bend to the machine, and that farms and worksites never will, so the machine has to learn to cope instead.

They start with berry harvesting. Delicate fruit, hidden under foliage, in variable outdoor conditions, which is arguably the hardest manipulation problem in agriculture. Europe is losing millions of farmers to retirement this decade and manual harvesting already accounts for the majority of berry production cost, so the labour problem is real and getting worse. Letters of intent are in place with growers in Southern Europe.

Berries are the wedge, not the destination. The methodology stays constant across markets: multimodal capture, simulation, hybrid training, field deployment. It generalises across fruit and veg and beyond that to any industry where skilled hands do work nobody has ever written down.

The founder's background is Formula 1 powertrain engineering and a high growth energy business. The company currently holds a place on a selective deep tech residency in Helsinki with a hardware lab and compute provided, with a live demo in front of investors in November and a priced seed round planned after.

THE ROLE, FOUNDING ML ENGINEER

You would own the learning stack. Motion retargeting is the critical path, taking captured human demonstrations and turning them into a policy that runs on a non human gripper.

The core problem is that a human hand has five fingers and twenty odd joints and the robot does not. You cannot replay the motion. You have to capture what the picker actually intended, approach, grip pressure, the twist, and translate that onto different kinematics without losing the skill.

The output is not a paper. It is a policy learned from human demonstration, running on real hardware, that picks the fruit without damaging it.

WHAT YOU WOULD BE BUILDING

• Egocentric hand tracking into a morphology neutral intermediate representation
• Retargeting onto robot hands with different kinematics to the human demonstrator
• Imitation policies with tactile in the control loop, not just in the training set
• Sim to real, sim for scale and real capture for contact ground truth, and you define where the crossing happens
• Engineering standards and direction for the learning stack as the team grows post raise

ESSENTIAL EXPERIENCE

• Shipped learned policies on real robot hardware. Sim only backgrounds are not a fit for this one
• Strong SE(3) geometry, kinematics and inverse kinematics
• Imitation learning, whether behaviour cloning, diffusion policies or VLA fine tuning
• Fluent PyTorch
• MuJoCo and or Isaac
• Comfortable owning a stack end to end with no one above you to check the work

DESIRABLE, NOT ESSENTIAL

• Dex retargeting, AnyTeleop or MANO pipelines
• EgoMimic or EgoVLA style work
• Tactile sensing
• LeRobot
• Dexterous manipulation research

If you have never done retargeting but you own the maths and have shipped on hardware, this is still worth a conversation. Retargeting is a frontier to be cracked and the fundamentals are what matter.

WHAT'S ON OFFER

• £20,000 for the three month contract period
• Converting to £80,000 plus 3 to 5 percent founding equity on a successful seed raise
• Founding engineer equity with standard vesting, credited from day one
• A path to CTO following the raise
• Full ownership of the learning stack, a hardware lab and captured data nobody else has
• Flights and accommodation covered for hardware lab access and the investor demo in Helsinki
• Fully remote in the UK

TO APPLY

Please apply via this advert or email directly. For further information, call . By applying, you consent to your data being processed in line with our privacy policy.