Robotics & AI Engineer — From Models to Real-World Robots

Astera Institute; California; Sur place; USD 120 000 - 180 000; Plein temps

Résumé du poste

Astera Institute is building a frontier robotics lab in Emeryville, California, seeking an engineer to take a learning-based robotics system from model development to reliable operation on a real robot. You will work across perception, policy learning, inference, and robot integration with researchers and hardware engineers.

This hands-on role focuses on manipulating capabilities, diagnosing failures on real robots, and improving software and models together in a collaborative lab environment.

Qualifications

• Experience in building or training models for robotics applications.
• Experience integrating perception systems with physical robots and calibration workflows.
• Strong software engineering practices for real-time robotic systems.
• Ability to diagnose failures across models, software, and hardware.

Responsabilités

• Develop models for robotic manipulation and perception-to-action pipelines.
• Connect sensors and model inference to robot control interfaces.
• Optimize inference latency and memory use on robot hardware.
• Implement robust policy execution and asynchronous inference.
• Diagnose failures across the full stack and build logging and evaluation tools.
• Package capabilities with clear interfaces and reproducible configurations.

Connaissances

ML engineering

Robotics integration

Systems debugging

Practical engineering

Technical collaboration

Robot learning

Inference optimization

Manipulation skills

Outils

PyTorch

ONNX

TensorRT

Description du poste

Astera Institute is building a frontier robotics lab in Emeryville, California, seeking an engineer to take a learning-based robotics system from model development to reliable operation on a real robot. You will work across perception, policy learning, inference, and robot integration with researchers and hardware engineers.

This hands-on role focuses on manipulating capabilities, diagnosing failures on real robots, and improving software and models together in a collaborative lab environment.