AI Research Engineer (Robot Learning)

Mimic Robotics•San Francisco, CA

1d•Onsite

About The Position

mimic is an early stage deep tech robotics & AI start-up based in San Francisco & Zurich and supported by leading VCs. We give industry workers a helping hand for tedious manual labor tasks and mitigate labor shortages with a versatile automation platform. Our automation solutions, driven by dexterous robotic hands and cutting edge AI trained on human observations, bring a new level of AI embodiment to the real world. As an AI Research Engineer (Robot Learning) you will drive frontier AI model development and data flywheel project management for dexterous robotic manipulation. You will shape all aspects of the development of end-to-end AI models for robotics, from large scale multi-task pre-training to specialized fine-tuning and post-training. You will also be responsible for the implementation and management of the data flywheel process, supervising and organizing task specification, data collection, model training and model evaluation in the real world. As an early employee, you will immediately co-lead AI engineering and science in close contact with the founders and early team.

Requirements

PhD in Computer Science, Data Science, Robotics or related field, or equivalent PhD-level experience in industry
3+ years of experience training generative models
4+ years of experience with developing and maintaining Python code
Strong empirical research abilities and intuitions
Project management / supervision experience
Fluent English speaker

Nice To Haves

Published deep learning and/or robot learning research in leading conferences (NeurIPS, CoRL, ICML, ICLR, etc.)
Experience in training large scale foundation models with distributed multi-GPU setups
Robotics and ROS2 experience
Experience with RL fine-tuning of generative models
Experience in open source software development

Responsibilities

Drive research and implementation for large scale, multi-task foundation AI models for robotics, quickly iterate over architecture design and parameters
Oversee model specialization for end use cases with fine-tuning and other post-training techniques
Drive the implementation and management of the data flywheel process, from data specification, to data collection, model training and real world evals
Write clean, maintainable python code using deep learning frameworks
Work closely with the Robotics and Hardware teams to ensure efficient and stable deployment on real world robots