Member of Technical Staff — Data Infrastructure

To achieve this breakthrough, we are building a Large Physics foundation Model (LPM) because physical systems, unlike text or images, are governed by verifiable cause and effect. We believe that scaling on physics will enable an understanding of causality required to predict and control physical systems, starting with weather.

We look for data engineers who are excited to tackle unsolved problems. Physical observations arrive continuously, in many formats, at a scale that dwarfs what is used to train today's LLMs. Your mission is to build the data platform underneath it all — the storage, compute, and loading systems that make every dataset cheap to ingest, fast to query, and immediately available to training.

Responsibilities

• Design and operate petabyte-scale storage: lakehouse architecture, file formats, and data layout optimized for both batch and real-time queries
• Own the shared compute and orchestration platform (e.g. Spark, Ray, workflow scheduling) that ingestion and research pipelines run on
• Optimize data strategy end to end from storage to loading, owning high-throughput data loading into training up to the tensor boundary
• Build systems for cataloging, deduplication, lineage, search, and reproducibility at every stage of the data lifecycle
• Implement the platform-level quality and monitoring tooling that data and research teams build their checks on
• Scale infrastructure to improve engineering velocity and ensure reliability, with monitoring and alerting to match
• Work across the full data lifecycle when the mission needs it — including building and operating ingestion pipelines for critical data sources directly

What we're looking for

• Demonstrated experience building large-scale data pipelines and distributed compute systems (e.g. Spark, Ray, Beam)
• Knowledge of state-of-the-art methods and tools for data ingestion, storage, and loading — including file formats and storage systems (e.g. Parquet, Zarr, Delta Lake) and how they impact performance and scalability
• Deep familiarity with cloud infrastructure, data lake architectures, and batch and streaming pipelines
• Understanding of how data loading throughput affects large-scale training, and experience optimizing it
• Owns deliverables end-to-end, from collecting and translating requirements to autonomously driving execution