Work

• A job that can crash, retry, or run twice — and still never lose an order

  Generating a public dataset export means producing several 300,000-record files at once. Doing that inside a web request is a timeout waiting to happen. The story of the async pipeline built to move it safely into the background — and make it survive crashes, duplicates, and downtime.

  Spring BootPostgreSQLRDF4JKubernetes

• From "grep every pod and hope" to one trace ID across a 7-service chain

  A record disappears somewhere inside a seven-service pipeline, and nobody can say where. The story of the baggage-tag trick that gave the system a memory — even the part running on a tool that can't read the tag.

  OpenTelemetry CollectorGrafanaTempoLokiPrometheusSpring BootKubernetes

• Cut filtered dataset search from 30+ minutes to sub-second at 25–50M records

  A library with tens of millions of books, and a librarian who took 30 minutes to answer a simple question. The story of teaching her a better catalog trick — without hiring a whole new library.

  MongoDBSpring BootKubernetesAngular / PrimeNGOpenSearch (evaluated)