Pipelines that don't go down.

I have spent about five years building data platforms that other teams trust to run the business. Most of that work lives where it is expensive to be wrong: claims pipelines in healthcare, fraud signals in insurance, and same day inventory and pricing in retail.

I gravitate to the parts people usually avoid. The 2am page, the schema that quietly drifted, the job that ran fine for a year until it didn't. That is where good engineering actually shows, and it is the part I enjoy most.

Day to day I work across three clouds with Spark, Kafka, Snowflake, Databricks, and dbt. I care about latency, cost, and data quality in roughly that order, and I write infrastructure as code so the next person never inherits a mystery.