Andrei Cabrera

Lead Senior Software Engineer  ·  Founder, LiftIQ

Platform engineering for the AI era.

I build the GraphQL platforms and agentic workflows that survive contact with production. Enterprise-scale by day; a bilingual iOS app and a public AI-skill lab, shipped solo.

Get LiftIQ on theApp Store

Lead Senior SWE  ·  Global hospitality platform  ·  Founder, LiftIQ

GitHub

Case studies

Recent work.

Client engagement

Domex Lawn Care: bilingual iOS app + cutover-ready web for a Brevard County service business

Live iOS app on the App Store. Next.js + Cloudflare marketing site staged behind the customer's GoDaddy site, with Resend email pipeline already verified on a send subdomain and the apex custom-domain wired on Cloudflare's side. Customer flips DNS when they're ready, not when I'm ready. EN/ES end-to-end, single-digit-$ infra.

iOSBilingual EN/ESNext.jsCloudflare WorkersResendManaged migrationClient engagement

What I work on

Engagement model.

A small slate per quarter, alongside LiftIQ. Three lanes of work, each with a defined scope.

Platforms

Federation & graph platform work

When your GraphQL graph crosses many teams, the technology stops being the problem and the org becomes the problem. I help platform teams ship subgraph patterns owners will copy, schema-check CI that catches breaks before the gateway, and deprecation playbooks teams will actually follow.

6 to 12 week scoped engagement · embedded with your platform team

AI engineering

Agentic features that survive production

Most "AI features" look brilliant in a demo and break under real load. I design evaluation harnesses, structured-output contracts, and MCP integrations so model-driven features ship, and stay shipped.

Discovery → eval scaffold → first feature shipped · 4 to 8 weeks

Advisory

Senior advisory for AI-first teams

For teams that don't need a full engagement but do need a senior set of eyes on architecture, evals, or platform decisions. Tight loop, narrow scope, no subcontracting.

Monthly retainer or per-engagement · by conversation

Lab

Working artifacts in agentic engineering.

Each one a small, real tool I built and use, recorded as a ~20-second demo. New ones land as I ship them.

liftiq-rn
Skills
Scaffolds a new LiftIQ feature (screen, hook, types, hook test, Maestro flow) in the project's exact house style: TypeScript-strict, Expo Router, Supabase RLS, Jest co-located.
architect-builder-tester
Skills
Multi-agent dev loop that ships full features end-to-end. Architect plans and surfaces ticket-vs-codebase conflicts. Builder executes (with bounded retry). Tester verifies on the real test suite. Two real PRs shipped tonight on the LiftIQ codebase, 7 conventional commits, 598/598 tests passing.
supabase-rls-mcp
MCP
Auth-aware MCP server on Cloudflare Workers. Claude Code queries my real LiftIQ data via 5 hard-scoped tools: no SQL exposed, service-role key encrypted in worker secrets, all queries gated to a single user UUID. 138 packages, zero vulnerabilities, 70-line custom SSE transport. Streamable HTTP MCP transport, deployed to supabase-rls-mcp.andrei-cabrera.workers.dev.

About

How I work.

I take ambiguous engineering work (platform-shaped, AI-shaped, product-shaped) and turn it into something the rest of the team can build on. The pieces I care about: evals that hold up under real load, schemas that resist drift, tests that mean something.

I work on a small slate at a time. Embedded with your team, not subcontracted, not running three projects in parallel. That discipline is the product.

  • 01Evals before features.
  • 02Defensible tech choices over fashionable ones.
  • 03Tests that mean something, not theater.
  • 04Small slate. Real depth. No subcontracting.

Endorsement

“Very experienced with GraphQL, React, Redux, and Redis. Helpful and descriptive. Doesn’t mind spending the time to explain certain technologies. I would definitely recommend Andrei to be a well-rounded asset to any team.”
Derek Sirola  ·  Former teammate  ·  June 2021

Contact

Bring me into the hard part of your roadmap.

Federation rollouts, agentic features, AI-assisted engineering workflows, evals you can actually trust. I take a small slate of contracts each quarter alongside LiftIQ.