The company is looking to shorten the feedback loop for metrics used in autonomous vehicles and delivery robots by building a platform that makes it straightforward to develop, run, and use metrics at scale. This involves handling complex compute graphs, large volumes of data, and optimizing storage for efficient inserts and fast reads.
Requirements
- Python in production, including async (e.g., asyncio, aiohttp, FastAPI).
- Strong SQL (JOINs, window functions, CTEs); ability to read plans and speed up slow queries.
- Data structures & algorithms—know when O(nlogn)O(n log n)O(nlogn) matters and choose the right structures.
- Experience with databases: PostgreSQL, ClickHouse; understanding OLAP vs. OLTP trade-offs.
- Workflow orchestration experience (Airflow / Argo / Prefect / Dagster—any is fine).
- Data libs & validation: NumPy, pandas, Pydantic (or equivalents).
- Containers & orchestration: Docker, Kubernetes.
Responsibilities
- Own the metrics platform: clear schemas, storage/layouts for efficient inserts and fast reads, simple versioning.
- Build and maintain the framework for writing/running metrics (interfaces, examples, local run, CI/compat checks).
- Create a test system for metrics and pipelines (unit / contract / regression on synthetic and sampled data).
- Operate the compute and storage paths in production; monitor, debug, and keep the system stable and cost-aware.
- Partner with metric authors and with development/analytics/QA to plan changes and land them safely.
- Design, develop, and operate the platform for metrics: data models and storage, scalable pipelines, and a developer-friendly framework for writing, testing, and shipping metrics.
- Choosing storage layouts and schemas that keep inserts efficient and reads fast.
Other
- Candidates are required to be authorized to work in the U.S.
- The employer is not offering relocation sponsorship.
- Remote work options are not available.
- Close collaboration with metric authors and with metric users—development, analytics, and QA.
- Ensuring reliable end-to-end delivery of metric results.
