Skip to content

Services

How I Can Help

My work sits where backend systems get hard: data from many external sources, work that runs in the background, rules that must hold for every tenant, and AI features in a regulated product. I design these systems so the important rules are enforced in one place and failures don't reach the people using them.

  • Backend Development

    You need a backend that holds up in production. I build APIs for web and mobile apps with Node.js or Python/FastAPI, on PostgreSQL, MongoDB, and Redis.

  • Multi-Tenant SaaS Backends

    Serving several customers from one product? I scope every query, job, and cache key to the right tenant so data never leaks between them.

  • Healthcare & EHR Integrations

    FHIR/HL7 and EHR vendor APIs, normalized into one model so each new hospital or vendor doesn't mean rewriting your product.

  • Background Jobs & Messaging

    Scheduled and event-driven work on Bull/BullMQ and Redis, with SMS and voice over Twilio that respects consent and quiet hours.

  • AI Features in Production

    RAG chatbots, LLM triage, and document generation added to your product with guardrails, fallbacks, and per-customer rollout.

  • Existing Backend Improvement

    A backend that's hard to change or quietly wrong? Audits, root-cause fixes, tests that prove concurrency and money logic, and safe expand/contract migrations.

What working together looks like

  • Root cause, not symptom

    I trace a reported bug back to the shared code path every caller goes through and fix it there.

  • Make critical rules structural

    Tenancy, consent, and quiet hours are enforced at one layer, not re-checked in every feature.

  • Decouple slow work

    Messaging and integrations run as idempotent background jobs, so failures retry safely and never block a workflow.

  • Roll out in phases

    Feature flags and per-tenant enablement, so changes can be rolled back without a redeploy.

  • Prove the safety claims

    Tests run on the real database engine, and mutation tests show a lock or guard actually prevents the failure.

  • Keep humans in the loop

    AI goes where it's safe. Where patient safety is involved, a person reviews the output.

Need a backend engineer who owns it end to end?

From architecture to production support, I take features from requirements to running software.