Donato Hernández | Full-Stack Developer

Full-Stack Developer

Full-Stack Developer · AI Integration · Backend Systems

I build full-stack products that ship.

Full-stack & backend developer behind Zaaby App and PitchPilot AI. Specialized in React Native, backend systems, and AI integration. Focused on shipping production-ready applications.

View Projects Let's Connect

01 · About

Behind the code

I started in logistics and systems administration, environments where downtime costs money and bad tooling costs more. That pushed me into building the tools myself: from a full-stack AI-powered iOS app shipped to the App Store, to real-time voice AI with sub-second latency running on Google Cloud. I care about owning the architecture end-to-end, writing code that's maintainable, and delivering products that solve real problems.

“What would this look like if it was simple, scalable, and actually shipped?”

Backend Engineering

AI Integration

Production Delivery

Data Architecture

Mobile Development

Workflow Automation

02 · Featured Work

Things I've shipped

Zaaby App

MVP Ready · TestFlight · Launching on App Store

Full-Stack DeveloperBackend ArchitectCo-founder

Pain point

People save great content from social media and never act on it.

AI-powered iOS productivity app that transforms saved links into clean summaries, action items, and structured takeaways. Built for people who save content but want to actually do something with it.

Technical Build

▸Full stack: React Native + Expo SDK 54, Supabase (PostgreSQL + Edge Functions in Deno), OpenAI GPT-4.1-mini, RevenueCat (3 subscription tiers)
▸Designed AI link processing engine: metadata extraction → state transition logic → structured output
▸Built onboarding flows with AI-driven personalization
▸Managed full build pipeline: EAS Build → TestFlight → App Store
▸Implemented RLS policies, GDPR/CCPA compliance, legal infrastructure

Impact & Results

✓Production-ready iOS app built from zero to TestFlight
✓AI processing pipeline handling multiple content types (articles, videos, TikToks via oEmbed)
✓Subscription system with 7-day free trial and 3 tiers
✓Security audit completed, zero critical vulnerabilities

What this proves

I can architect and ship a complete AI-powered mobile product: backend, data, payments, compliance, and deployment. Solo.

Visit Landing Page

PitchPilot AI

Shipped · March 2026

Full-Stack DeveloperBackend EngineerCo-founder

Pain point

Founders practice pitches without real-time, role-specific feedback.

Real-time AI pitch practice platform that simulates specific audiences (investors, teachers, clients) via bidirectional voice conversation powered by Google Gemini Live API.

Technical Build

▸Backend: Node.js + TypeScript, WebSockets as audio proxy between browser and Gemini Live API
▸Deployed on Google Cloud Run, frontend on Vercel
▸Built phase monitoring system to detect AI failure points, drastically reducing system crashes
▸Transformed audio sessions into detailed written reports with performance metrics

Impact & Results

✓Real-time bidirectional voice AI with sub-second latency
✓Intelligent phase detection reducing failure rate significantly
✓Post-session reports with business advice and action plans
✓Successfully migrated infrastructure from shared to personal GCP account

What this proves

I can build real-time, production-grade AI systems with complex infrastructure. Not just CRUD apps.

Coming Soon

Linaria

Live · Production

Full-Stack DeveloperAI Agent Architecture

Pain point

Small dev teams waste time on manual standups and don't catch blockers until it's too late.

An AI executive memory agent for development teams. Connects to your team's GitHub repo, analyzes commits, PRs, and recent activity, then automatically generates daily standups, detects blockers, alerts on risks, generates changelogs, and displays team velocity in a visual dashboard. Features a self-improvement loop where the agent evaluates its own summary quality using Arize Phoenix for observability.

Technical Build

▸Python + Google ADK with Gemini 2.5 Flash as core LLM
▸FastAPI backend deployed on Google Cloud Run
▸React + TypeScript + Tailwind frontend on Vercel
▸Arize Phoenix MCP integration for tracing and observability
▸Self-improvement loop: agent evaluates and scores its own outputs
▸Multi-step planning engine for complex queries
▸PyGithub for real-time repo analysis (commits, PRs, issues)

Impact & Results

✓10 production features: auto standup, blocker detection, changelog generation, PR review queue, team velocity chart, risk alerts, repo selector, multi-step planning, self-evaluation, Phoenix observability
✓Automated daily standups replacing manual team updates
✓Real-time blocker detection across PRs and issues
✓Self-improving agent that gets better with each interaction

What this proves

I can architect and ship a complete AI agent system, from LLM orchestration and observability to production deployment on Google Cloud.

Coming Soon

03 · Values

What I believe in

Ship over perfect

Working software beats perfect plans every time.

Systems over patches

If a problem repeats, I build something to stop it.

AI as a tool, not a crutch

I use AI to go faster, not to skip understanding.

Clarity in code

Readable, maintainable code is a form of respect for your team.

Remote-first mindset

Async communication, documentation, ownership.

Talent over location

The best work happens when you hire for skill, not timezone.

04 · Skills

Stack & tools

Backend & Data

Node.js · TypeScript
Python · FastAPI
PostgreSQL · Supabase
Edge Functions (Deno)
REST APIs
Data modeling & RLS

Frontend & Mobile

React Native · Expo
Next.js · React
JavaScript · TypeScript
EAS Build · TestFlight

AI & Automation

OpenAI API (GPT-4.1-mini)
Google Gemini Live API
Google ADK (Agent Development Kit)
Prompt engineering
AI workflow integration
LLM API integration

Tools

Git

GitHub

Jira

Supabase

Vercel

Google Cloud

RevenueCat

Arize Phoenix

Figma

Claude

Windsurf

Git

GitHub

Jira

Supabase

Vercel

Google Cloud

RevenueCat

Arize Phoenix

Figma

Claude

Windsurf

“AI should accelerate understanding, not replace it. I use it to build faster, but I know why every line works.”

05 · Contact

Let's Work Together

If your team needs a developer who ships full-stack products, integrates AI thoughtfully, and works well async, let's talk.

I'm open to remote full-stack roles, especially teams building with AI.

Email Me

manueldonato9921@gmail.com

Connect with me

SMS / WhatsApp

+52 647 122 9788