Projects⁰⁵

Products I built solo, end to end: from the database to the shipped app. The source is private, so here are the real screens running.

Lot feed with opportunity score, current bid and FIPE comparison

Lot detail: score, risk, potential profit and margin

Mobile app · ML · scraping

Auction triage with computer vision

Mobile app (Android/iOS) that aggregates vehicle auctions from several auction houses and ranks opportunities. A scraper collects the lots, an OCR + NLP pipeline extracts risks from the PDF notices, and a computer-vision microservice (CLIP, zero-shot, no GPU) classifies damage in the vehicle photos. The scoring engine combines market (FIPE) price, risk and potential margin. Built for low operating cost, with no paid AI APIs.

My role

Solo, end to end: Laravel API, Playwright scraper, OCR and computer-vision pipelines, scoring engine, React Native app and CI/CD with automated deploy.

Stack

React Native
Expo
Laravel
PostgreSQL
Redis
Playwright
Tesseract OCR
CLIP / PyTorch
Docker
GitHub Actions

Screenshots of the app running locally, with lots collected by its own scraper.

Private sourceSee in the feed

Skill library with categories and filters

SaaS marketplace · payments

Skill marketplace for AI agents

Platform where developers publish and monetize reusable “skills” for AI agents. High-performance back-end with Laravel + Octane, payments and seller payouts via Stripe Connect, and a Next.js front-end. Containerized monorepo with automated deploy.

My role

Full-stack + infra, solo: marketplace modeling, API, split-payment integration, front-end and Docker environment.

Stack

Laravel 12
PHP 8.4
Octane / Swoole
PostgreSQL
Redis
Stripe Connect
Next.js 15
React 19
Docker

Running locally with demo data.

Private sourceSee in the feed

Map of prioritized hospitals with one hospital's panel open

Data engineering · geospatial

In progress

Public-health data pipeline

Python pipeline that cross-references 100% public DATASUS data (hospital production, structure, accreditations and funding ceilings) to locate and prioritize philanthropic hospitals that are candidates for SUS revenue recovery. It includes a composite scoring model, facility geolocation and an interactive map.

My role

Concept and solo development: pipeline architecture, open-data ingestion, prioritization model and visualization.

Stack

Python
ETL
DATASUS / SIGTAP
Pandas
Leaflet

Map generated by the pipeline from public DATASUS data.

Private sourceSee in the feed

Dashboard with the month's sales, revenue, profit and stock

Product catalog with stock, expiry and status

Full-stack · Shopify integration

E-commerce operations dashboard

Internal dashboard to run an e-commerce operation: products, kits, stock and expiry control, sales, expenses, a partner portal and spreadsheet imports. Automatic sync with Shopify via API and webhooks.

My role

Full-stack, solo: database modeling, application, Shopify integration and an authenticated partner portal.

Stack

Next.js
React
TypeScript
Supabase / PostgreSQL
Shopify API
Tailwind CSS

Running locally with made-up data, no real business data.

Private sourceSee in the feed

Extension panel on a product page: profit, ROI, demand and restrictions

Chrome MV3 · micro-SaaS

Browser extension for product research

Chrome extension (Manifest V3) that injects an analysis layer onto Amazon product pages: data scraping, FBA profitability calculation, sales estimates from the sales rank and category-restriction checks. Freemium model with usage limits and payments.

My role

Solo: extension architecture, scraper, profitability calculator and freemium payment gating.

Stack

JavaScript
Chrome Extension MV3
Web scraping
ExtensionPay

The panel is the extension's real code, running on a test product page.

Private sourceSee in the feed