Member of Technical Staff, Robot Learning

present of AI; San Francisco (CA); On-site; USD 180,000 - 240,000; Full time

Job summary

Darwin develops general-purpose humanoid robots designed to learn physical manipulation tasks through real-world experience. The role focuses on building models and policies that let robots learn from demonstrations and real-world data, balancing research with deployment.

The candidate should bring deep learning foundations and a strong ML background, with a passion for experiments, data-driven evaluation, and applying ideas to real hardware. This is an on-site position in San Francisco.

Qualifications

• Strong machine learning foundations and depth in at least one ML area.
• Careful thinking about experiments and data.
• Drive to carry ideas from research through deployment.

Responsibilities

• Build models and policies that enable humanoid robots to learn physical tasks from demonstrations and real-world experience.
• Develop data collection and training pipelines for robot learning.
• Test and validate policies on actual robots with dexterous manipulation tasks.
• Explore new learning architectures and evaluate performance in real hardware environments.

Skills

Deep learning foundations

Manipulation learning

Job description

Member of Technical Staff, Robot Learning

The role asks for a strong machine learning researcher or engineer to build models and policies that let humanoid robots learn physical tasks from demonstrations and real world experience. The work spans dexterous manipulation, new learning architectures, data collection and training pipelines, and testing policies on actual robots. A strong candidate brings deep learning foundations, depth in at least one machine learning area, careful thinking about experiments and data, and the drive to carry ideas from research through deployment.

How it fits the AI era

This role sits at the center of the AI era, where learned policies replace hand coded instructions for physical work. AI changes how robots pick up assembly and tool use skills, shifting effort toward data collection, training, and evaluation on real hardware. What stays human is judgment about which problems matter, how to design experiments, and how to judge whether a policy is truly reliable.

How it builds AI experience

The posting does not require prior AI experience beyond strong machine learning foundations, so the role itself is a way to build it. A candidate can prepare by training models on demonstration data, studying imitation and reinforcement learning for control, and getting hands on experience with evaluation on real systems. Because Darwin focuses on humanoid robots learning manipulation from demonstrations, practice with robot learning pipelines and physical testing would be directly relevant.

Worth showing

• Deep learning foundations
• Manipulation learning

This is our own read, written from the employer's posting and the company's Spotlight. The employer's posting is the source of truth for the role.

About Darwin

Darwin develops general-purpose humanoid robots designed to learn physical manipulation tasks through real-world experience, per the company.