Machine Learning Engineer (Robotics)

Location: Prague, Czech Republic; hybrid work

Company

hirify.global is a Central European e-grocery company building an autonomous grocery operation with AI assistants, logistics automation, computer vision, and robotics.

What you will do

• Define camera, mounting, calibration, synchronization, and episode-quality specifications for human-video capture in fulfilment centres.
• Own hand-pose ground truth and address occlusion, work gloves, cold warehouse conditions, and other real-world data challenges.
• Build an evaluation harness that determines whether captured footage is suitable for robot-policy training.
• Post-train open robot-learning models on the collected data and run initial task evaluations.
• Read and reproduce published research, validate its claims in warehouse conditions, and guide capture decisions before significant spending.
• Act as the primary machine-learning voice on a newly created team alongside an operations lead and data engineer.

Requirements

• Strong PyTorch experience and practical computer vision knowledge involving 3D geometry.
• Experience with camera calibration, pose estimation, and SLAM fundamentals, including debugging extrinsic calibration.
• Ability to read and reproduce published papers and understand the requirements of current robot-learning policy classes.
• Experience training on self-collected data and diagnosing dataset failure modes.
• Ability to make independent technical decisions as the sole machine-learning specialist and direct AI coding agents while maintaining quality.
• Experience with hand-pose estimation, egocentric video, multi-camera rigs, time synchronization, robot-learning model post-training, or ROS 2 is particularly relevant.

Culture & Benefits

• Opportunity to build the physical-AI side of an autonomous grocer from the beginning.
• Ownership of the capture specification and first training runs.
• Access to off-the-shelf hardware and a managed cloud environment.
• Work alongside a small, newly created team focused on turning warehouse recordings into a training asset.
• Engineers use Claude Code and Devin to support harness development, data plumbing, and reproduction scripts.

Hiring process

• Instead of a cover letter, select a published human-video-to-robot-policy pipeline and explain where it would fail in a warehouse where workers wear gloves.