Intern, Robot Learning

Baseline Robotics; Zürich; On-site; CHF 30,000 - 36,000; Full time

Job summary

Baseline Robotics in Zurich, Switzerland, seeks a role to advance policy learning for robotics. You will train policies on demonstration data and benchmarks, building datasets and evaluating across task variations.

The position requires Python and PyTorch experience, familiarity with diffusion policies or VLA models, and RL policy learning from demonstrations. On-site 5 days a week, 3–6 month contract, CHF 3,000 monthly.

Qualifications

• Experience training neural networks end-to-end.
• Fluency in Python and PyTorch.
• Exposure to diffusion policies, transformers, or vision-language-action (VLA) models.
• Familiarity with reinforcement learning (PPO, SAC) or policy learning on demonstration data.
• Experience with GPU-accelerated simulation tools (Isaac Lab, Newton, mjlab, Genesis).

Responsibilities

• Train policies on demonstration data and run benchmarks on real hardware.
• Build and characterise datasets behind a benchmark, including failure cases and coverage gaps.
• Run policy evaluations across task variations and analyse where success rates break down.
• Reproduce published manipulation results.

Skills

Neural nets training

Diffusion policies/transformers/VLA

Reinforcement learning

Tools

Python

PyTorch

Isaac Lab

Newton

mjlab

Genesis

Job description

About Baseline Robotics

Open source revolutionised how we write software, but its impact on robotics stays limited due to expensive hardware accessible only to well‑funded labs. We run that hardware as shared infrastructure for the broader community, billed only for active compute. For a given task, policies are evaluated under similar conditions, giving researchers common ground for fair comparisons.

In this role you will

• Train policies on demonstration data and run benchmarks on real hardware.
• Build and characterise datasets behind a benchmark, including failure cases and coverage gaps.
• Run policy evaluations across task variations and analyse where success rates break down.
• Reproduce published manipulation results.

What we hope you'll bring

• Experience training neural networks end-to-end.
• Fluency in Python and PyTorch.
• Exposure to diffusion policies, transformers, or vision-language-action (VLA) models.
• Familiarity with reinforcement learning (PPO, SAC) or policy learning on demonstration data.
• Experience with GPU-accelerated simulation tools (Isaac Lab, Newton, mjlab, Genesis).

Details

• Location: Zurich, Switzerland, on-site 5 days a week.
• Duration: 3-6 months.
• Monthly salary: CHF 3,000.

Deadline to apply: None. Applications will be reviewed on a rolling basis.