Rerun is building a new multimodal data stack for robotics and spatial AI, aiming to transform the physical-world economy by providing the necessary data infrastructure and tools. The company is seeking a Robotics ML Engineer to enhance their open-source project for teams developing AI-first robotics products.
Requirements
- Experience with MLOps and the surrounding systems needed to get data from a robot to GPUs (e.g. Linux, Ray, file formats, Pytorch dataloaders, NVIDIA DALI)
- Strong Python skills
- basic skills in C++ or Rust
- Broad enough knowledge of ML, 3D geometry, robotics and related fields to build useful projects and communicate effectively with customers and the community
Responsibilities
- Building examples and reference projects that use both Rerun's open source and commercial data infrastructure for embodied AI use cases.
- Data pipelines for robotics imitation learning
- Data cleaning workflows
- End-to-end physical robot to learned behavior setup
- Simulation for evaluation and data generation
- Using those projects as a basis for creating blogs, tutorials, videos, and other educational materials to help our customers and community
- Use all the above to feed back learnings into the core Rerun product, sometimes by directly fixing things yourself
Other
- An ownership mindset - we're all doing this together!
- A helpful attitude and a curious mind
- A track record of cool projects and public communication (blog, youtube, great READMEs)
- We meet up in person for a week roughly once a quarter
- value agency and helpfulness highly, and expect everyone to take broad responsibility for what they build
