I like the unglamorous parts of data.

I'm David Mwandairo, a data engineer and data scientist. The work I enjoy most sits between where data arrives and where decisions come out: the ingestion that survives a flaky API, the deduplication that makes the grain honest, the tests that catch the drift before the dashboard does.

My portfolio reflects that. One project lands raw market JSON in an immutable bronze layer and walks it through dbt to point-in-time-correct marts. Another streams every OLTP change out of a microfinance core through Debezium and Kafka into ClickHouse, with Grafana watching each hop. A third takes a churn problem end to end: EDA, leakage-safe features, model selection, then a tested FastAPI service behind Docker.

I care about the parts reviewers ask about: idempotency, auditability, latency you can measure, and numbers you can trace to a query.