Synthium is building the data engine for Embodied AI.

We capture human cognition through interactive simulations and transform multimodal behavior into structured datasets used to train next-generation robotics foundation models.

We are looking for an Embodied AI & Simulation Engineering Intern to support the development of our simulation, embodiment, and data generation stack.

Role Description

• Work closely with the founders and research partners on systems bridging simulation, teleoperation, embodiment, and large-scale dataset generation

• Contribute to the robotics simulation architecture powering Synthium’s data engine

• Help prototype and build the foundations of our robotics engineering stack

What You’ll Work On

• Contribute to the simulation stack across Isaac Sim / Isaac Lab and MuJoCo

• Support development of cloud-scalable pipelines for multimodal dataset generation

• Help design human-in-the-loop and teleoperation systems for data capture

• Assist with imitation learning and reinforcement learning data workflows

• Contribute to systems spanning control, perception, policy learning, and evaluation

• Help improve simulation fidelity, reproducibility, and AI-robotics workflows

What We’re Looking For

• Strong interest in robotics, simulation, or Embodied AI

• Experience with Python

• Familiarity with robotics simulation environments (Isaac, MuJoCo, etc.) is a plus

• Familiarity with PyTorch or TensorFlow

• Curious, hands-on, and comfortable operating in ambiguity

• Excited to work in a fast-moving startup environment

Nice to Have

• Interest in 3D graphics, ML systems, data pipelines, or cloud infrastructure

• Experience with humanoid robotics

• ROS2 experience

• Experience with teleoperation systems

Why This Role Matters

This role sits at the core of Synthium’s platform: building the infrastructure that turns human behavior into training data for robots.

The systems you design will directly shape how robots learn, reason, and interact with the world.