I own the systems your business can't afford to break.

Moved a production Rails 6.1 application off Heroku onto self-hosted infrastructure with Docker and Kamal — one week of work, around 30 hours, replacing both production and staging. Annual hosting fell ~97% while application memory grew 2.3×, which works out to roughly 75× more RAM for every dollar spent.

Turned a five-year-old single-tenant Rails application into a platform that safely holds many companies — 99 tables brought under tenant isolation, ~105 controllers and 45 Sidekiq jobs made tenant-safe. Isolation is enforced at the request boundary rather than by a model-level default_scope, and proven by around 50 per-controller scope specs plus a query guard that fails the build on any unscoped read.

Built an event-driven Amazon SP-API integration for a catalogue of roughly 40,000 SKUs across 2 marketplaces. Listing-issue notifications arrive through EventBridge and SNS on a webhook endpoint, severity routes them so a suppression and a warning are not treated alike, and 5–6 remediation workflows keyed on Amazon's own issue code correct the listing — pricing, missing catalogue data, restrictions — then verify the correction actually took. An unmapped issue code is escalated to a person rather than guessed at.