Frontend developer · Lima, Peru

I own the outcome, from messy problem to production, and I direct AI across the build.

A decade-plus of hand-written software backs every call I make directing AI. I'd rather show you the work than tell you about it.

Let's work together See the case studies

At a glance

years building software
16
personal projects built on my own
11
teams led, from 2 to 7 people
5
Selected work

Three projects: AI-orchestrated, AI in the product, hand-built

All projects
Pandacorp

Pandacorp

AI-orchestrated

An AI software factory I built and run: 3 projects taken through it, and this site built and tested in 5 days.

Next.jsTypeScriptAI orchestration
Read the case study
Job Research

Job Research

AI in the product

A conversational job search for frontline workers: I led the app and built the pipeline that ingests +1000 offers a day.

Next.jsVercel AI SDKInngest
Read the case study
Alohub

Alohub

Hand-built

One editing engine instead of a panel per section type, and 75% less development time.

ReactNext.jsGoogle Maps
Read the case study
The honest version

How I really use AI

I'm not an AI engineer and I won't pretend to be one. Nothing ships until the tests, the types and a real browser render all pass. I direct AI across design, code and review.

See a full product I built this way: Pandacorp

AI does the execution. I own every call that matters.

Design: I set the direction, AI explores the options.
Code: AI writes the groundwork, I own the architecture.
Review: AI flags the first pass, I make the final call.
The stack

The tools I build with

Frontend
ReactNext.jsTypeScriptTailwind CSSshadcn/uiReact Hook FormZodTanStack Query
AI in the build
Claude CodeCodexVercel AI SDK
Backend & data
Node.jsPrismaPostgreSQLFirebasebetter-auth
Quality & delivery
PlaywrightVitestSentryPostHogVercelDockerGitHub Actions

Got a problem worth solving?

That's the part I like best. Tell me what you're building.

Let's work together