Machine Learning Intern, Autonomy

Gravis Robotics is a startup that turns heavy construction machines into intelligent and autonomous robots. Our unique combination of learning-based automation and augmented remote control lets one operator safely conduct a fleet of machines in a gamified environment—from anywhere in the world. Our team has over a decade of academic experience honing the cutting edge of large-scale robotics, and is rapidly growing to bring that expertise into a trillion-dollar industry through active deployments with market leaders. We are looking for passionate, skilled interns with a background in machine learning to join our team—and to actively contribute to the development and deployment of extraordinary construction robots. The ideal candidate should be self-motivated, capable of working autonomously in a team and have a strong desire to solve exciting, challenging, and applied problems.

As part of the Autonomy team, you will focus on designing, testing, and benchmarking machine learning models, as well as conducting ablation studies to understand their performance and limitations. The insights generated through your work will help improve these models and support downstream applications, e.g. control policy synthesis, ultimately contributing to faster and more accurate machine digging.

Responsibilities

• Design, test, and deploy novel ML models for autonomous heavy machinery
• Benchmark and analyze model performance, including conducting ablation studies to evaluate key design choices.
• Help define performance metrics, validation methodologies, and explore anomaly detection methods to understand the limitations of each architecture and which kind of data is needed for robust performance of the ML models.
• Collaborate closely with the rest of the teams to improve the reliability and scalability of the ML pipeline.

Qualifications

• Proficiency with Python, and Git.
• Familiarity with popular deep learning libraries (PyTorch, etc.).
• Experience with data analysis, ML optimization, and hyperparameter tuning.
• Strong analytical and problem-solving skills, with the ability to interpret experimental results and draw sound conclusions.

The following experience is considered a plus:

• Model-based reinforcement learning.
• Large-scale robotics simulation environments, such as NVIDIA Isaac Sim.
• Robot Operating System, including ROS or ROS 2.
• C++.