Robotics / Physical AI Research Engineer

About this role

Innodata seeks a Robotics/Physical AI Research Engineer to bridge data operations and robotics foundation model requirements. You'll design data specifications, guide collection efforts across multiple modalities, and validate that data design choices measurably improve model performance for robot learning applications.

What you'll do

• Translate robotics foundation model requirements into concrete data specifications covering modalities, action representations, and annotation schemas
• Structure and enrich existing and incoming data to maximize value for training and evaluating robot foundation models
• Guide robotics data collection across capture modalities including motion capture, teleoperation, and synthetic generation
• Conduct experiments validating data design choices through fine-tuning and evaluating foundation models
• Develop evaluation and benchmarking methodology for robotics data grounded in measurable criteria
• Collaborate with capture lab, annotation teams, and synthetic data pipeline to operationalize data specifications

What they're looking for

• Robot learning and robotics machine learning (imitation learning, reinforcement learning, manipulation)
• PyTorch and strong Python programming
• Vision-language-action models and world models
• Robotics data formats (LeRobot, RLDS, Open X-Embodiment)
• Fine-tuning large vision-language models with HuggingFace transformers and PEFT
• Robotics simulation tools (NVIDIA Isaac Sim, Isaac Lab, Omniverse)
• Teleoperation and egocentric data collection
• Technical communication with both expert and non-expert audiences

Benefits

• Salary range: $180,000 - $300,000 USD per year
• Hybrid role with frequent on-site presence in Saddle Brook, NJ
• Mid to senior level opportunity
• Work on cutting-edge physical AI and robotics foundation models
• Collaborate with robotics research organizations