Robot Autonomy Engineer

DeepRec.ai; München; On-site; EUR 120,000 - 160,000; Full time

Job summary

DeepRec.ai is hiring a Staff Robot Learning Engineer in Germany. The role focuses on training and improving policies for humanoid robots, developing data pipelines, and running experiments on physical hardware to tackle sim-to-real transfer.

You will work across imitation learning, reinforcement learning, and vision-language-action models to turn advances into reliable robotic capabilities. The position requires hands-on engineering experience with real robots, strong Python and PyTorch skills

Qualifications

• 2+ years of professional experience training real robot policies using RL, IL or learning from demonstration.
• Experience deploying learned policies on physical robots.
• Strong Python and PyTorch skills, with good C++.
• Experience with robot data collection, simulation and sim-to-real transfer.
• Solid understanding of robot control, action representations and closed-loop policy execution.

Responsibilities

• Train and improve robot-learning policies for humanoids.
• Develop data pipelines for robot learning projects.
• Run experiments on physical hardware and study sim-to-real transfer challenges.
• Contribute to building robust, real-world robotic capabilities.

Skills

Reinforcement learning

Imitation learning

Learning from demonstration

Python

PyTorch

C++

Robot data collection

Sim-to-real transfer

Robot control

Closed-loop policy execution

Tools

Python

C++

Job description

Staff Robot Learning Engineer | Germany | Full-time

You’ve already trained and deployed learning-based policies on real robots. Now you can help bring the next generation of humanoid robotics into the real world.

You’ll join an ambitious robotics company developing general-purpose humanoids, building robot-learning systems that work beyond the simulator. You’ll train and improve policies, develop data pipelines, run experiments on physical hardware and tackle the challenges of reliable sim-to-real transfer.

This is a hands‑on engineering role for someone who combines strong machine-learning fundamentals with practical robotics experience. You’ll work across imitation learning, reinforcement learning and emerging vision‑language‑action models, helping turn promising approaches into robust robotic capabilities.

You’ll need:

• Minimum 2+ years professional experience training real robot policies using reinforcement learning, imitation learning or learning from demonstration.
• Proven experience deploying learned policies on physical robots.
• Strong Python and PyTorch skills, with good C++.
• Experience with robot data collection, simulation and sim‑to‑real transfer.
• A solid understanding of robot control, action representations and closed‑loop policy execution.

Experience with humanoid manipulation, teleoperation, diffusion policies, vision‑language‑action models or real‑time action chunking would be a strong advantage.

If you want to turn advances in robot learning into capable, reliable humanoid robots, let’s chat!