Machine Learning Engineer VLA Training and Evaluation

You will build reusable training and evaluation workflows for robot foundation models. You will fine-tune vision-language-action models, develop policy-training loops and benchmarks, conduct sim-to-real transfer studies, and turn published research into extensible baselines.

Responsibilities

• Build reusable model-agnostic training and evaluation workflows
• Implement fine-tuning for pretrained vision-language-action models
• Build training loops for action models, diffusion policies, and VLM/VLA heads
• Build benchmarks, success metrics, and failure-analysis templates
• Collaborate on sim-to-real transfer studies
• Turn published methods into working baselines and templates

Requirements

• Experience writing and debugging training loops in PyTorch or JAX
• Experience fine-tuning pretrained models
• Understanding of imitation-learned robot policies
• Experience designing evaluations and ablations
• Experience with experiment tracking, versioning, and checkpoint management

Benefits

• Corporate Benefits Program covering health, mobility, and learning for 100€ net per month
• Free drinks and fruit
• Regular company events