Forward Deployed Researcher (Robot Learning)

MicroAGI, Inc.; München; On-site; EUR 50,000 - 70,000; Full time

Job summary

MicroAGI, Inc. is seeking an early-career researcher to train and deploy neural policies for embodied tasks across real and simulated data. You will work on data, training, evaluation, and deployment pipelines for robotic systems and collaborate with deployment engineers at customer sites.

The role emphasizes rapid learning from university labs, theses, internships, or research positions, with a focus on robust real-world performance and transfer to physical robots.

Qualifications

• Experience training modern neural-network-based robotic policies through a thesis, internship, or research project.
• Familiarity with imitation learning, reinforcement learning, vision-language-action models, or related robot-learning methods.
• Strong Python and hands-on experience with PyTorch, JAX, or an equivalent framework.
• Experience running experiments on real robots or in a robotics simulator.
• Good understanding of machine learning fundamentals and experimental design.
• Evidence of strong technical work such as a thesis, paper, or open-source contribution.

Responsibilities

• Train neural policies for manipulation and other embodied tasks using real and simulated data.
• Build data, training, evaluation, and deployment pipelines for robotic policies.
• Run experiments in simulation and transfer successful policies onto real robots.
• Collect demonstrations and deployment data, diagnose failure modes, and improve policies iteratively.
• Evaluate models for robustness, generalisation, latency, and real-world task success.
• Work alongside deployment engineers at customer sites to adapt policies to new environments and workflows.
• Turn research ideas into systems that operate reliably outside the lab.

Skills

Python

PyTorch

Reinforcement learning

Imitation learning

Robotics

Research experience

Experiment design

Deployment to real robots

Learning quickly

Education

Bachelor's degree in a related field

Tools

PyTorch

JAX

MuJoCo

Isaac Sim

ManiSkill

Job description

The next ten years of AI will not be won in software. They will be won in the physical world: in factories, hospitals, kitchens, fields, and homes. The companies that own that data will own the century.

MicroAGI is building it. We are the data layer for physical AI.

You will train and deploy modern neural robotic policies for real customer tasks. This is an early-career role for someone whose strongest experience may come from a university laboratory, thesis, internship, or recent research position. We care about the depth of your work and your ability to learn quickly, not the number of years on your CV.

What You Will Do

• Train neural policies for manipulation and other embodied tasks using real and simulated data.
• Build data, training, evaluation, and deployment pipelines for robotic policies.
• Run experiments in simulation and transfer successful policies onto real robots.
• Collect demonstrations and deployment data, diagnose failure modes, and improve policies iteratively.
• Evaluate models for robustness, generalisation, latency, and real-world task success.
• Work alongside deployment engineers at customer sites to adapt policies to new environments and workflows.
• Turn research ideas into systems that operate reliably outside the lab.

Requirements

• Experience training modern neural-network-based robotic policies through a thesis, university laboratory, internship, research project, or early professional role.
• Familiarity with imitation learning, reinforcement learning, vision-language-action models, or related robot-learning methods.
• Strong Python and hands-on experience with PyTorch, JAX, or an equivalent framework.
• Experience running experiments on real robots or in a robotics simulator.
• Good understanding of machine learning fundamentals and experimental design.
• Evidence of strong technical work, such as a thesis, paper, research project, open-source contribution, or working robotic system.
• Comfortable moving between research code and physical hardware.
• Willingness to travel to deployment sites.

Nice to Have

• Experience with manipulation, dexterous control, teleoperation, or whole-body policies.
• Familiarity with Isaac Sim, MuJoCo, ManiSkill, or similar environments.
• Experience with diffusion policies, transformers, multimodal models, or vision-language-action systems.
• Publications or workshop papers in robotics, computer vision, or machine learning.
• Experience collecting or curating robot-training data.