Machine Learning Scientist (Robotics & Physical AI)

Empathy Talent; New York (NY); On-site; USD 120,000 - 180,000; Full time

Benefits offered by this job

Equity

Job summary

Empathy Talent in New York City is seeking a versatile ML engineer to advance robot learning and deploy intelligent systems in production warehouses. The role blends research with production, requiring deep hands-on coding and collaboration with robotics and deployment teams.

You will train models, optimize them for edge hardware, build data pipelines, and evaluate new approaches, contributing to a fast-moving startup that ships impactful automation technology.

Qualifications

• Experience deploying ML models in production environments.
• Proficient in PyTorch, TensorFlow, or Ray.
• Strong production-grade coding ability.
• Experience across the ML lifecycle from research to deployment.
• Ability to work autonomously in a fast-moving startup.
• Generalist mindset across ML and robotics stack.
• Passion for robotics, AI, and physical automation.

Responsibilities

• Improve how quickly and reliably robotic systems learn and perform new tasks.
• Train vision-language-action models for robotic grasping, manipulation, and control using datasets.
• Evaluate new robot learning approaches and apply promising methods to production systems.
• Build data pipelines for data collected by deployed robots.
• Optimize neural networks for large-scale training and edge inference.
• Integrate trained models into the production robotics stack.
• Collaborate with robotics and deployment teams to scale performance across environments.

Skills

ML in production

PyTorch

TensorFlow

Ray

Production code

ML lifecycle

Embedded systems

Education

Master's degree in CS/ML/Robotics

Tools

ROS2

Simulation environments

Edge hardware

Job description

Join an early-stage robotics company developing general-purpose automation systems that are already operating in production environments across U.S. warehouses. The team is building intelligent robots that can be deployed quickly, learn new tasks, and continuously improve using data collected from real-world operations.

This is a highly ambitious, fast-paced environment where scientific depth, engineering quality, and execution all matter. The team values versatile builders who enjoy working across research and production and want to help advance robot learning, physical AI, and accessible automation.

Location

New York City (On-site)

What You'll Do

• Improve how quickly and reliably robotic systems learn and perform new tasks.
• Train vision-language-action (VLA) models for robotic grasping, manipulation, and control using simulated and real-world datasets.
• Evaluate emerging approaches in robot learning, including reinforcement learning and imitation learning, and apply promising methods to production systems.
• Build pipelines for collecting, organizing, and curating data generated by deployed robots.
• Optimize neural network models for large-scale training and low-latency inference on edge hardware.
• Integrate trained models into the production robotics stack.
• Partner with robotics and deployment teams to scale model performance across different warehouse environments and operating conditions.

What We're Looking For

• Strong experience training, deploying, and maintaining machine learning models in production environments.
• Deep knowledge of modern deep learning methods and frameworks such as PyTorch, TensorFlow, or Ray.
• Ability to develop high-quality production code and contribute effectively within a software engineering team.
• Experience working across the machine learning lifecycle, from research and experimentation through deployment and ongoing improvement.
• Comfortable taking ownership and setting priorities in a fast-moving, high-autonomy startup environment.
• A practical, generalist mindset and willingness to work across multiple layers of the ML and robotics stack.
• Passion for robotics, artificial intelligence, and building systems that operate in the physical world.

Preferred Qualifications

• Hands-on experience applying machine learning to robotics or working directly with robotic hardware.
• Background in real-time ML inference, reinforcement learning, imitation learning, or sim-to-real transfer.
• Experience training vision-language-action models or other multimodal architectures.
• Familiarity with robotics sensors, sensor calibration, and real-world data collection.
• Knowledge of Python, C++, ROS2, simulation environments, or low-latency control systems.
• Master’s degree in Computer Science, Machine Learning, Robotics, or a related field.

Compensation

• Equity

This is an excellent opportunity for someone who wants to advance robot learning and physical AI while working directly with production systems, real-world data, and robotic technology that is already delivering value in operational environments.