Senior AI Research Engineer - Foundation Models | Senior AI Research Engineer (m/w/d) Foundation Models

Agile Robots SE

München, Bayern, Deutschland
Published Mar 4, 2026
Full-time
Permanent

Job Summary

As a Senior AI Research Engineer at Agile Robots, you will spearhead the development of learning-based robot policies and multimodal foundation models for embodied intelligence. Your daily work involves designing vision- and language-conditioned policies using imitation learning, diffusion models, and transformer architectures. You will take full ownership of model-level decisions, defining architectures and training objectives that translate perception into real-world robot behavior. A key part of the role is analyzing policy behavior under uncertainty and collaborating with cross-functional robotics and perception teams to ensure seamless integration. This position is particularly attractive for researchers who want to see their work move beyond fine-tuning into end-to-end learning pipelines that control physical systems. Based in Munich, you will join a highly international team of experts working at the cutting edge of force-sensing and image-processing technology, bridging the gap between advanced AI and physical robotics in a fast-growing, well-funded environment.

Required Skills

Education

Master’s or PhD in Computer Science, Robotics, AI, or a related technical field.

Experience

  • Professional experience in designing and training learning-based policies for robotics using imitation or reinforcement learning
  • Hands-on experience adapting or designing transformer-based or vision-language-action models for embodied tasks
  • Demonstrated experience in ML engineering including implementation of training loops and experimentation workflows
  • Experience in model-level ownership and debugging policy failures in real-world or semi-real environments
  • Experience with large-scale training and GPU clusters is preferred

Languages

Not specified

Additional

  • The role is based in Munich, Germany. Candidates must be able to work in a collaborative, interdisciplinary environment with 60+ nationalities. The position involves end-to-end ownership of learning pipelines from data assumptions through real-world validation.