Datamentors

ABOUT US

We’re a robotics startup building an agnostic orchestration platform that allows companies to manage, monitor, and control multiple robots and robotic systems from a single interface. We’re also developing custom Vision-Language-Action (VLA) models that enable natural-language reasoning and autonomy across diverse robot types and industries. Our mission is to make robotics more accessible, intelligent, and scalable across sectors from hospitality and healthcare to defense and sports facility management. Think of it as an operating system for any robot, with an expanding library of intelligent capabilities. We are self-funded and currently building the v1 platform with our first pilot partners in the energy, hospitality and sports sectors.

ROLE OVERVIEW

Join the AI team building the learning brain of the platform. You train and fine-tune the VLA models that enable natural-language reasoning and autonomy across diverse robot types — from hospitality to defence. You own the data flywheel from raw teleop demonstrations through to deployed models.

TECH STACK

• PyTorch
• VLA / ACT / Diffusion Policy
• CUDA
• LeRobot / HF

WHAT YOU’LL DO

• Train and fine-tune VLA models that enable natural-language control and autonomy across diverse robot types and industries
• Build and improve video-prediction and world-model pipelines that improve robot planning
• Develop vision models for robot perception feeding real-time control across all embodiments on the platform
• Own the data flywheel: collection, curation, augmentation and systematic evaluation pipelines
• Work hands-on with the robot fleet to close the train–deploy–iterate loop and measure real task success rates

WHAT WE’RE LOOKING FOR

• 5+ years ML/AI with hands-on training depth — you have written the training loop, not just called a wrapper
• Depth in VLAs, video models or large vision models; robot learning, imitation learning or RL is a strong plus
• Expert PyTorch and multi-GPU training; comfortable debugging training divergence and eval-to-deployment gaps
• Experience building data curation and evaluation pipelines for model training
• Ability and enthusiasm to work alongside real robot hardware, not just in simulation

Nice to have

• Experience with ACT, Diffusion Policy or π0-style policy architectures
• Isaac Sim or MuJoCo experience for synthetic data generation

WHY JOIN US

• This is a rare opportunity to help design and build the foundational technology for a robotics platform from the ground up
• Competitive salary
• Relocation support and housing assistance for Madeira-based roles; CINM free-zone benefits
• Small, senior, hands-on team — everyone owns a meaningful slice of the product from day one