What are the responsibilities and job description for the Robotics Manipulation / Physical AI Internship position at Autonomique?
About Autonomique
Autonomique is a startup building at the forefront of Physical AI: making robots do real work in the real world. Our intelligence framework solves the core challenges of autonomy—perception, reasoning, and dexterity—to empower reliable operations for the world’s most demanding industries. We're already deploying bimanual robots in production.
The Internship
We are looking for exceptional, hands-on interns excited about pushing the state of the art in robot manipulation, learning-based control, and real-world deployment. You will work directly with the founding and senior robotics team on problems that sit at the intersection of learning, control, and systems engineering.
This is not a “research-only” or “observer” role—your work will be integrated into our core stack and deployed on physical robots operating under tight latency, safety, and reliability constraints.
What You’ll Do
- Deploy Robust Manipulation Algorithms: Design and implement algorithms for complex bimanual tasks (pick/place, insertion, tool use) that don't just work in the lab but perform reliably in dynamic industrial environments.
- Develop learning-based robot behaviors using approaches such as Diffusion Policy, imitation learning, and reinforcement learning to improve performance, robustness, and recovery.
- Contribute to end-to-end robot autonomy systems that connect perception, semantic task understanding, and low-level robot control.
- Build Production-Grade Systems: Contribute production-level C and Python code to our core robotics stack, prioritizing testing, maintainability, and real-time performance.
Must Haves
- Educational Background: Currently pursuing a MS or PhD in Robotics, Computer Science, or a related field.
- Manipulation Experience: Prior work or projects involving robotic arms, dexterous manipulation, inverse kinematics, motion planning, or grasping.
- Learning-Based Robotics: Experience with imitation learning, reinforcement learning, or diffusion/sequence models applied to control or robotics.
- Vision Skills: Familiarity with computer vision and pose estimation frameworks
- Software: Strong proficiency in Python; solid C experience is a plus.
- ML Frameworks: Deep understanding of PyTorch.
- OS: Ubuntu or related Linux.
Bonus Points
- Experience with ROS / ROS2 in real systems.
- Hands-on work with physical robots (beyond pure simulation).
- A strong “builder” mindset—you’re comfortable debugging hardware issues, calibrating sensors, or diagnosing flaky real-world behavior.
- First-author publications at top-tier venues (ICRA, CoRL, RSS, CVPR, NeurIPS).
- Publicly available code, projects, or research (e.g., GitHub).
Logistics
- Location: Menlo Park, CA or Montreal, QC (On-site)
- Duration: 4 to 6-months.
- Start Date: Flexible.