SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
Physical Intelligence is building general-purpose AI systems for the physical world, developing foundation models and learning algorithms to power robots and physically-actuated devices. As an ML Infrastructure Engineer focused on Data Systems, you will design and operate the data infrastructure that powers large-scale robot learning at scale.
You'll work at the intersection of distributed systems, storage, and machine learning infrastructure. Your systems will sit directly between raw data sources and training/evaluation pipelines, enabling the company to move faster while maintaining performance, correctness, and reliability.
Key responsibilities include: designing and building high-throughput data ingestion and processing pipelines that validate, transform, and featurize raw multimodal data; operating large-scale batch and streaming workflows over massive datasets; designing object storage layouts, metadata systems, and efficient access patterns; building systems for backfills, dataset rebuilds, garbage collection, and large-scale transformations; optimizing dataloaders, sharding, prefetching, caching, and throughput to reduce latency from data arrival to model training; building scalable metadata stores for datasets, annotations, and training artifacts; moving petabytes efficiently across clusters and environments; implementing observability, validation, and guardrails to prevent silent data regressions; and collaborating cross-functionally with researchers, engineers, and roboticists to translate evolving data needs into robust systems.
You should have strong software engineering fundamentals, experience building distributed systems or large-scale data pipelines, comfort reasoning about performance/memory/I/O/storage efficiency, familiarity with batch and/or streaming processing systems, and experience with object storage systems and data format tradeoffs. An ownership mindset—designing, building, operating, and iterating on systems end-to-end—is essential. You'll enjoy working closely with researchers and unblocking fast-moving projects.
Bonus experience includes large ML training pipelines or dataloading systems, knowledge of columnar or custom data formats, hands-on experience with systems like ClickHouse, Ray, Flink, or Spark, operating petabyte-scale datasets, and debugging performance bottlenecks in data-heavy systems.
