Research Scientist, Robotics & World Models

Innodata Inc.

4d•$160,000 - $185,000

About The Position

Innodata is seeking a Research Scientist to lead the science of data for robotics foundation models. This role involves partnering directly with customers and frontier labs to define data specifications, guide data collection and synthesis, and develop evaluation methodologies. The goal is to enable the responsible advancement of AI by providing high-quality data and evaluation frameworks for next-generation robotics foundation models, including vision-language-action models, world models, and manipulation and locomotion policies. The scientist will be responsible for ensuring that the data collected and synthesized is relevant and transferable to real-world robotic applications.

Requirements

Roughly 4+ years of hands-on industry experience in robot learning or robotics ML.
A Bachelor's degree in computer science, electrical engineering, robotics, or a related technical field is required.
Trained and evaluated robot policies yourself (manipulation or locomotion via imitation learning or reinforcement learning) with strong PyTorch fundamentals.
Experience curating, filtering, and weighting robot data across embodiments and sensors.
Fluency in robotics data formats and standards (LeRobot dataset format, RLDS, Open X-Embodiment), along with common motion and sensor formats.
Hands-on experience with simulation and synthetic data (NVIDIA Isaac Sim, Isaac Lab, Omniverse, MuJoCo) including domain randomization, system identification, and sim-to-real transfer.
Experience with teleoperation or egocentric data collection.
Comfort adapting VLM backbones for control and fine-tuning large VLA models with the modern toolchain (HuggingFace transformers, PEFT).
A track record recognized in the field: first-author publications or strong open-source contributions at venues such as CoRL, ICRA, IROS, RSS, or NeurIPS.
Ability to work directly with research scientists at customer organizations and explain data and modeling decisions clearly to both expert and non-expert audiences.
A rigorous, reproducible approach to experiments and documentation.

Nice To Haves

An advanced degree (MS or PhD) in a relevant field is preferred.
Interest or hands-on experience in responsible-AI evaluation and red-teaming for embodied systems, such as safety and robustness testing.

Responsibilities

Define how Innodata designs, structures, and evaluates data for robot foundation models, and validate these choices experimentally.
Translate the requirements of robotics foundation models (vision-language-action models, world models, manipulation and locomotion policies) into concrete data specifications, including modalities, action representations, sampling, annotation schemas, and evaluation criteria.
Decide what data is worth capturing in the real world versus generating in simulation, and curate and weight training mixes across heterogeneous robot datasets.
Guide data collection across various modalities (motion capture, egocentric, exocentric, teleoperation, multi-sensor) and synthetic pipelines.
Build evaluation and benchmarking methodology that predicts real-world transfer, including world-model evaluation and sim-to-real gap closure.
Run experiments to fine-tune and evaluate foundation models on Innodata data, demonstrating the impact of data decisions on model improvement.
Design adversarial and long-horizon evaluations to identify policy and world model failure modes and use these to improve data collection.
Publish benchmarks, methodology, and papers to advance the field and build trust with partners.
Collaborate with capture labs, annotation teams, and synthetic data pipelines to translate specifications into operational plans.