Software engineer · Full stack · Applied AI

I build whole products, from interface to deployment.

Architecture, interface, API, and deployment. Also the layer AI pilots are missing to survive real users, and the debugging of systems already in production and failing.

Genaro Melgar — Software engineer · Full stack · Applied AI

What I do

Four fronts I work on.

01

Full stack product development

React, Next.js, and TypeScript up front; Node.js and Python behind. Architecture, interface, API, and deployment.

02

AI agents in production

From prototype to a system that holds real traffic: guardrails, cost control, idempotent execution, and evals.

03

Backend, APIs, and integrations

Data models that survive growth, APIs that don't break when versioned, and third-party integrations: payments, video, calendars, CRMs.

04

Debugging systems at scale

Bugs that only surface in production, on real data across thousands of accounts. Proven root cause, the fix, and repair of the data already damaged.

Experience

Teams and products I've worked on.

Arketa

Full stack software engineer

Product work on a platform running bookings, payments, and memberships for thousands of fitness and wellness studios. Interface, backend, and third-party integrations.

Root AI

Software engineer · Applied AI

AI agents that operate real CRM workflows, and the backend underneath them.

Hybridge Education

Software engineer

A conversational admissions agent, plus backend and infrastructure work on the institution's platform.

Projects

Product I've built end to end.

Multi-branch operations ERP

Point of sale, inventory with transfers and shrinkage, staff management, and daily closeout generation. A monorepo with its own component library, published as a package and consumed by the other apps.

  • Next.js
  • React
  • TypeScript
  • Tailwind CSS
  • Supabase
  • TanStack Query
  • Zod
  • Sentry

BeGrown

Clinical nutrition platform

A tool for private practices: sessions, custom measurements, menus, and patient progress tracking. A monorepo with domain types and validators shared across apps.

  • Next.js
  • React
  • TypeScript
  • Turborepo
  • Supabase
  • PostgreSQL
  • Zod

Betin

Prediction model and bot

A match-prediction pipeline built around Elo ratings and momentum features, publishing its results to Telegram on a schedule.

  • Python
  • APScheduler
  • Docker

Technologies

What I work with, by area.

Frontend

  • React
  • Next.js
  • React Native
  • TypeScript
  • Tailwind CSS
  • TanStack Query
  • Zustand

Backend

  • Node.js
  • Python
  • FastAPI
  • Express
  • REST
  • GraphQL
  • Zod

Data

  • PostgreSQL
  • Supabase
  • MongoDB
  • Firestore
  • BigQuery

Infrastructure

  • Docker
  • AWS ECS
  • AWS RDS
  • Google Cloud
  • Netlify
  • Sentry

Applied AI

  • OpenAI Agents SDK
  • Guardrails
  • Evals
  • Idempotent execution

Integrations

  • Stripe
  • Zoom
  • Calendars
  • CRMs

Contact

Tell me what you need built or what's breaking. Call or email me directly — you reach me, not a form.