Ashans
Khadka

I led the one-month move from a compliance vendor to an in-house deletion API — cutting six-figure annual vendor costs.

Regulatory deletion (GDPR/CCPA) reaches into every store we keep customer data in. We ran it through a third-party orchestrator: expensive, and a core legal obligation living outside our control. I designed and built the in-house replacement.

The rules keep moving — new stores, new retention windows, new regulation — so I built it around a domain layer with schema versioning. The business logic stays put while the storage underneath can change without a rewrite.

What I'd do differentlyI let one system's document shape stand in for an explicit contract between the two systems. A later change dropped identifier fields and broke a downstream table. I'd define, validate, and version that data contract on day one.

Zero-downtime MongoDB engine upgrades across seven environments.

I completed every major version jump — 4.4 → 5 → 6 → 7 → 8 — with my team across five production and two pre-production environments. Standby-cluster safety nets, a validated rollback path, and maintenance windows coordinated across regions.

Somewhere in there I became the person the org asks about MongoDB. I try to make that scale as a strength, not a bottleneck — reusable runbooks and pulling teammates into the real upgrades so the knowledge spreads.

Reliable MongoDB-to-lakehouse change data capture.

I helped establish Terragrunt as a reusable layer for provisioning Databricks infrastructure, then maintained the CDC path from MongoDB through Debezium and Strimzi on Kubernetes to managed Kafka, S3, and Delta Lake.

To demonstrate the pipeline was dependable, I built validation showing 100% agreement between Kafka events and landing data, then presented the result to the wider engineering organization.