Guillermo Casanova

Senior software engineer with 5+ years across startups, fintech, e-commerce, and AI-native products. I build agent infrastructure, fast web and mobile interfaces, and observable, cost-aware systems.

01 · Engineering focus

Four things I keep measurable.

Load

Fast initial load

Image budgets, lean bundles, prefetching, and removing work before users feel it.

UX

Predictable user flows

Flows that stay predictable under real data, with clear states and accessible feedback.

DX

Developer experience

Typed APIs, reusable primitives, fast local feedback, and CI that helps teams move without guessing.

Signal

Observable quality

Tests, evals, observability, and production metrics so regressions are visible early.

02 · What I build

Product systems that ship and stay shipped.

01

AI-native product systems

Agent workflows, tool calling, structured outputs, eval harnesses, and observability loops that make model behavior measurable before it reaches production.

02

Product infrastructure

Frontend, mobile, backend-for-frontend, caching, search, and deployment pipelines when the product needs speed without splitting ownership.

03

Measured delivery

Small, verifiable increments: profile the bottleneck, reduce the moving parts, automate the check, and keep the business metric visible.

03 · Current stack

Tools I reach for first.

AI

  • Vercel AI SDK
  • Anthropic
  • OpenAI
  • Gemini
  • MCP
  • Claude Code
  • Cursor
  • Evals

Product

  • React
  • Next.js
  • TanStack Start
  • TanStack Query
  • Shadcn/ui
  • Zustand
  • TypeScript

Mobile

  • React Native
  • Expo
  • EAS
  • MMKV
  • FlashList

Platform

  • Node.js
  • NestJS
  • Bun
  • Supabase
  • Redis
  • GCP
  • AWS
  • Vercel
  • Terraform
  • Docker
  • GitHub Actions

Quality

  • Playwright
  • Vitest

Observability

  • PostHog
  • Sentry
04 · How I work

Four working rules I don't break.

  1. Own the outcome across the product surface, not only the ticket boundary.
  2. Treat performance, cost, and quality as numbers to observe, not opinions to debate.
  3. Automate the repetitive, frictionless workflows let the team spend its judgment on outcomes, not toil.
  4. Use AI and automation for repeatable engineering work, then verify the result with tests and production signals.