Attis is a well-funded Series A climate/AI startup seeking a Staff Software Engineer focused on Data Infrastructure and Real-Time Pipelines. In this deeply hands-on role, you will design, build, and run production systems that serve as the data and compute backbone for an AI Earth System aimed at weather and climate analysis.
Responsibilities:
- Build data ingestion & pipelines
- High-throughput ingest for real-time & batch data (models, satellite, radar, reanalysis, obs)
- Turn messy external feeds into analysis-ready, cloud-optimized datasets
- Design bursty GPU & low-latency workflows
- Architect systems that can spin up large GPU fleets briefly, then scale back down
- Keep end-to-end latency low and costs sane
- Productionize inference and evaluation with research/ML teams
- Own cloud-native systems
- Make pragmatic choices across AWS/GCP, Modal/serverless, containers, batch
- Own monitoring, alerting, SLOs, runbooks, and failure-mode thinking
- Enable the rest of the team
- Build tools/abstractions so researchers and data scientists can run large workflows safely
Requirements:
- Raise the bar on code quality, testing, CI/CD, and observability
- Senior / Staff‑level IC: Typically 7+ years as a hands‑on SWE or data/infra engineer
- You've taken systems from idea → design → production → operations
- Real‑time / large‑scale data: Built and run high‑throughput, low‑latency pipelines or big batch systems
- Designed for reliability, observability, and cost at scale
- Python in production: Strong modern Python for data/infra
- Comfortable turning prototypes into robust, maintainable services
- Cloud‑native (AWS and/or GCP): Deep experience with managed services to build scalable, resilient systems
- Containers + serverless/batch environments (Docker, K8s, Modal or similar)
- Workflow orchestration: Designed and operated complex workflows in Airflow / Dagster / Prefect or similar
- High autonomy: You can ramp quickly on a complex domain
- You're comfortable in a small, distributed, senior team with little hand‑holding
- Domain‑relevant large data: Weather, climate, environmental, satellite, radar, or geospatial data
- Formats like GRIB, NetCDF, Zarr, Parquet, etc
- Earth system / NWP exposure: Worked near numerical weather prediction or large scientific models
- GPU / HPC‑style workloads: Operating GPU‑accelerated or large batch compute (HPC, Slurm, Ray, K8s, Modal)
- Research / open‑science environments: Translating research code into robust production systems (academia, labs, open‑source)
- Early‑stage startup experience: Wearing multiple hats, making pragmatic trade‑offs, and shipping under uncertainty