Data & Distributed-Systems Engineer

Dave Anaya

I'm a data engineer. I founded CatInCloud Labs to build fault-tolerant data systems end to end, from the hosts to the pipelines, and to study how distributed systems fail.

From Data Engineering to Systems

For years, I worked as a Data Engineer. I optimized large pipelines (BigQuery, Snowflake, Airflow) and built real-time dashboards (React).

The pipelines were the part everyone could see. The part I cared about was underneath them: what a system does when an API rate-limits it, a host restarts or the network drops packets, and whether it recovers on its own.

CatInCloud Labs is where I do that work: its own hosts, its own pipelines, and research on how distributed systems fail. The research is pre-registered before the results, and the engineering decisions are written up in the log.

Some of it is open source. The packages that came out of this work are on the Systems page, with what each one does and how it is kept honest.

Calm Systems

Complexity belongs in the problem, not the infrastructure. Everything here is built on a "calm architecture" philosophy that keeps the work reproducible and cost-efficient:

  • Reliability Is Evidence: If the pipeline breaks at 2 AM, the data is corrupted, and every result built on it is invalid. I prioritize idempotency and self-healing code to ensure data integrity.
  • FinOps Is Architecture: I design for efficiency (e.g., local vectorization), so the budget goes to the work, not to cloud overhead.
  • Private by Default: Catalyst's host is reached over Tailscale and its disk is encrypted with LUKS, and the scheduled workflows and the Slack agent that call the dex service hold no warehouse credential. That is where a system here starts, not a layer added at the end.

The Stack

These are the tools the systems here are built and tested with today.

Computing Python, Polars, DuckDB
Storage Parquet, Google Cloud Storage, BigQuery, Postgres
Orchestration Dagster, dbt Core, Docker
Infrastructure Hetzner Cloud, Tailscale, Caddy, Terraform, SOPS
Observability Prometheus, Grafana, Loki