As a Machine Learning Engineer on the Atlas team, you will join a world-class team of engineers and scientists focused on creating next-generation Machine Learning foundational models for perception to power behavior development. Our team is interested in workflows that support training, fine-tuning, evaluation and deployment of foundational models that use embodied data. In this role, you will be responsible for proposing, evaluating and integrating new workflows and architectures to create these perception models that power Atlas to solve challenging manipulation tasks. Your contributions will help us build new robot technologies for Atlas and other R&D efforts at Boston Dynamics.
How you will make an impact:
- Design and implement scalable machine learning pipelines for embodied data
- Develop software pipelines to support both cloud and on-prem GPU cluster workloads
- Design systems to mine for valuable data and organize large in-house datasets
- Create real and synthetic datasets for training ML models
- Develop tools to query and gain insights into data, training, and testing performance metrics
- Build software to refine the models through hyperparameter optimization and structured experimentation
- Regularly present your research within the Atlas team and in wider company meetings
We are looking for:
- 3+ years experience developing large-scale ML infrastructure, tools, pipelines
- Advanced knowledge in Python programming
- Familiarity with popular ML frameworks - pytorch, tensorflow
- Experience with ML deployment (ex. Triton, ONNX)
- Experience with database systems for scalable ML dataset management
- Experience with docker and job orchestration for cloud and on-premise GPU clusters
Nice to have:
- PhD in Computer Science, Machine Learning, Robotics, or a related field
- Experience working with and contributing to large codebases
Read Full Description