Oswaldo Ortiz ● Senior Software Engineer — NYC
I build systems that survive the real world.
From embedded kiosks on clinic floors to HIPAA-compliant data platforms and AI agents in production — owned end to end.
I'm a senior software engineer at a HIPAA-regulated med-spa group in New York, building the software the business runs on. Because I also run IT, I live with everything I ship — so durability isn't a slogan, it's self-preservation.
I care about graceful degradation, observable behavior, and the kind of boring reliability that lets a business forget its infrastructure exists.
- 0 kiosks in production
- 0 recurring costs eliminated
- 0 appointments/day through my software
- 0 staff using my tools daily
-
A fleet of 16+ custom photo kiosks deployed across studios — designed from the hardware up: Raspberry Pi 5, a purpose-built Linux image, and a native UI rewritten in C with Raylib for instant responsiveness. Over-the-air updates patch every booth at once, and a Prometheus–Grafana–Loki stack over a Tailscale mesh watches each kiosk — camera status, photo counts, thermal limits, even SD-card wear prediction — so failures are diagnosed in minutes, not site visits.
-
A Slack-native IT assistant giving employees 24/7 self-service support — deflecting roughly 20% of tickets by walking staff through fixes and filtering out non-issues before they reach a human. A Claude Sonnet orchestrator delegates to Haiku sub-agents for RAG distillation, response validation, and ticket enrichment, grounded in a pgvector knowledge base — with a human-in-the-loop approval gate and per-feature flags so any piece can degrade gracefully on its own. It superseded the Slack ticketing system I'd built the year before.
-
In-house prescription-fulfillment tooling that replaced a $60K-per-year vendor. Built on a HIPAA-handling Postgres database with address validation at order entry, it kept patient data under our own controls, diversified the pharmacy supply chain, and committed the business to a two-business-day ship SLA.
-
Per-vial inventory tracking for neuromodulators across 15 locations: scanning workflows at the fridge, disposal-rate forecasting for replenishment, weighted point systems for product variants, and a full changelog so every unit is auditable from delivery to disposal.
-
A revenue-ranked lead queue replacing an unstructured Salesforce workflow for a private medical practice. SQL scoring signals — speed-to-lead, booking intent, unconverted quotes, reactivation windows — rank by collectible revenue, with an AI layer planned for judgment-call narratives on top.
Let's build something durable.
● Open to contract engineering work