Research Scientist (Physical AI)

Location: Zurich, Switzerland; hybrid workplace

Company

hirify.global develops robotic systems and is building foundational physical AI models from large-scale deployment data generated by robots operating in production environments.

What you will do

• Design Transformer- and Diffusion-based foundation model architectures, objectives, and training curricula for multimodal robotic data.
• Develop scalable data mixtures and sampling strategies using vision, action, and state data from robot deployments.
• Run ablations to investigate scaling laws, data quality, optimization dynamics, and large-model failure modes.
• Develop fine-tuning, imitation learning, reinforcement learning, distillation, and curriculum learning methods for real-world robot adaptation.
• Improve robot reliability in factory environments, including robustness to out-of-distribution edge cases.
• Build physical evaluation setups and benchmarks, then use real-world results to guide research and model development.

Requirements

• Deep research and practical experience across machine learning, systems engineering, and physical robotics.
• Experience designing, training, and fine-tuning large-scale deep learning architectures, including VLMs, VLAs, RL, RLHF, or imitation learning.
• Strong fundamentals in PyTorch or JAX and the ability to debug across the full stack.
• Hands-on comfort with robotic hardware and understanding of perception, controls, and state estimation.
• Experience deploying and validating policies on real hardware, with a rigorous approach to evaluation and failure analysis.
• English-speaking working environment.

Culture & Benefits

• Work with real robots and production deployment data on practical physical AI problems.
• Combine machine learning research with engineering and direct real-world testing.
• Collaborate with experienced researchers and engineers from major technology companies, startups, and robotics organizations.
• Flexible working hours.
• Relocation package for the Zurich-based role.

Hiring process

• Phone screen with the hiring manager.
• Half-day onsite process covering cultural fit and deep-dive technical interviews.
• Final decision within 2–3 days after the onsite interview, with detailed feedback provided.