Founding Robot Learning Engineer (Remote/London) — Equity

Upplands Brohus; Greater London; Remote; GBP 135,000 - 165,000; Full time

Benefits offered by this job

Equity ownership

Remote-friendly London

Founding engineer equity

Job summary

Upplands Brohus is seeking a Founding Robot Learning Engineer to join a YC-backed physical AI and robotics startup in London with remote options. You will bridge human demonstration data and policy performance, deciding what datasets to collect and how to represent them.

Reporting to the Engineering Lead, you will build task segmentation, data loaders, and policy training experiments using egocentric and UMI datasets to prove data improves robot task success, with equity ownership over policy

Qualifications

• Background in robot learning engineering, embodied AI, or robotics ML.
• Proven experience preparing training data and evaluating policies on real or simulated tasks.
• Hands-on work linking dataset decisions to policy improvements and avoiding data leakage.

Responsibilities

• Turn long recordings into task intervals and define frame-locked labels.
• Deploy AI models to propose task boundaries and route uncertain results for review.
• Build data loaders and transformations for video, labels and motion trajectories.
• Fine-tune and evaluate robot policies on human demonstration data with controlled tests.
• Translate evaluation insights into precise data recommendations for perception and capture.

Skills

Robot learning

Python

PyTorch

Dataset tooling

Policy evaluation

Temporal segmentation

Tools

Data loaders

Job description

Upplands Brohus is seeking a Founding Robot Learning Engineer to join a YC-backed physical AI and robotics startup in London with remote options. You will bridge human demonstration data and policy performance, deciding what datasets to collect and how to represent them.

Reporting to the Engineering Lead, you will build task segmentation, data loaders, and policy training experiments using egocentric and UMI datasets to prove data improves robot task success, with equity ownership over policy