Robot Learning Engineer - Manipulation

About this role

Applied Intuition seeks a Robot Learning Engineer to develop and deploy manipulation policies for industrial robots, working across the full pipeline from task definition through real-world validation. You'll train policies using modern approaches like vision-language-action models and diffusion policies, then optimize and evaluate them on physical hardware to meet customer performance targets.

What you'll do

• Train, fine-tune, and deploy manipulation policies from task definition through real-robot evaluation and deployment
• Work with large pretrained models and compact task-specific policies, selecting appropriate approaches for different manipulation tasks
• Develop repeatable recipes for industrial tasks including pick-and-place, bimanual handling, and contact-rich assembly
• Deploy policies on edge compute and validate observation processing, action interfaces, and control timing
• Diagnose policy failures and convert failures into improved data, models, and evaluation strategies
• Measure and optimize customer-focused metrics such as success rate, cycle time, and intervention rate

What they're looking for

• Python and PyTorch for training and deployment code
• Imitation learning and modern policy families (vision-language-action models, diffusion policies, action-chunking transformers)
• Robot kinematics, coordinate frames, and camera calibration
• Integration of learned policies with low-level robot control
• Experimental diagnosis and debugging across data, sensing, and execution
• ROS 2 (nice to have)
• Edge inference optimization on NVIDIA Jetson or GPU systems (nice to have)
• Reinforcement learning and policy transfer across platforms (nice to have)