S
Skild AI builds foundation models and control systems for robots using PyTorch, TensorFlow, JAX, and ROS 2. The stack spans simulation (Blender, Three.js, React Three Fiber), real-time telemetry (WebSockets, WebRTC), and cloud infrastructure (AWS, GCP, Azure), indicating a pipeline from training through deployment. Engineering dominates hiring (34 of 56 open roles), with heavy emphasis on senior IC roles—reflecting the research-intensive phase of training large-scale robot models. Active projects target model training infrastructure, data pipelines, and animation-to-simulation workflows; pain points cluster around scaling both data annotation and RL training, plus deploying prototypes to production reliability.
Tech Stack 67 technologies
Core StackC++ Python PyTorch TensorFlow AWS Node.js JavaScript TypeScript React MongoDB Docker Figma Java Linux SolidWorks ROS 2 JAX GCP Azure WebSockets WebRTC WebXR Socket.IO SQL React Three Fiber Three.js Blender ROS Rhino KeyShot+37 more
What Skild AI Is Building
◆Challenges
- Maintaining fleet reliability
- Scaling data annotation operations
- Improving serviceability
- Scaling training infrastructure
- Monitoring working capital metrics
- Scaling up reinforcement learning model training
- Robust deployment
- Scaling research from prototypes to deployable solutions
- Optimizing ml models for real-world robotics
- Improving robotics system reliability
▲Active Projects
- Robot control and monitoring systems
- Model training infrastructure
- Develop new algorithms for training general-purpose robot foundation models
- Dashboard automation
- Animation pipelines for humanoid characters
- Building large scale data pipelines
- Embedded control system improvement
- Integration of animation data into real-time simulations
- Motion capture pipelines
- Training pipeline development
Hiring Activity
Steady55 roles · 25 in 30d
Department
Seniority
Company intelligence
