Building products end-to-end — from user interfaces to the data pipeline.
Storefront, admin panel and API for Sophia Laços, a Brazilian handmade hair-bow atelier. A pnpm + Turborepo monorepo: a React 19 storefront with catalog, cart and WhatsApp checkout, a TanStack Router admin with Google sign-in, an order pipeline and product photo upload, and a Hono + tRPC Cloudflare Worker over D1 (Drizzle ORM), R2 for images and Stripe Checkout. Deployed to Cloudflare Workers and released by version tag.
Corporate website for MICROTECH, a specialized Apple repair shop, featuring an interactive iPhone repair quote builder that estimates service cost and hands off to WhatsApp. Built with React 19, TypeScript, Tailwind v4 + shadcn/ui and Zustand, deployed to Cloudflare Workers.
Self-hosted, on-prem production cluster running real factory operations at Grendene. Five-node Docker Swarm (all managers) over distributed GlusterFS storage, fronted by Traefik v3 with a private Docker Registry and Portainer. Runs PostgreSQL 18 + pgvector, an Airflow (CeleryExecutor) ETL pipeline feeding a datalake, a local AI stack (Ollama + Open WebUI + Valkey with pgvector embeddings), and 30+ Streamlit/nginx services including 13 factory-monitoring dashboards.
Full-stack web app for managing production activities in Grendene's tooling department. Built end-to-end: a NestJS + Prisma REST API with JWT auth and role-based access control, plus a React 19 SPA (React Router, TanStack Query for server state, Zustand for client state, Zod-validated React Hook Form) on a shadcn/ui interface. Containerized with Docker.
A suite of custom Python tool functions for the self-hosted 'Grendene AI' (Open WebUI + Ollama) assistant: live web fetching, CSV/JSON parsing, Portuguese text processing and tokenization, datetime helpers, and Ollama/vector-store health checks. Backed by a Docker Swarm AI stack with GPU-accelerated Ollama, RAG over pgvector (nomic-embed-text embeddings), and Valkey-managed sessions.
A cloud-agnostic, Databricks-inspired Data Lakehouse running entirely on Docker Compose. MinIO (S3) object storage holds Apache Iceberg tables, exposed through an Apache Polaris REST catalog and queried by Trino. Spark/PySpark and dbt drive a Stage → Bronze → Silver → Gold medallion flow orchestrated by Airflow, fed by Airbyte batch ingestion and Debezium + Kafka CDC. Schema is gated by data contracts, quality by Great Expectations checkpoints (with a quarantine lane), lineage and governance by OpenLineage → OpenMetadata, and observability by Prometheus + Grafana.
End-to-end IoT monitoring: Rust firmware on an Arduino Uno reads sensors over serial, a Python bridge streams the data, and a React dashboard renders it in real time — embedded systems, hardware integration, and web front end as one stack.
Educational web app that simulates an ALU through an interactive alarm system built from logic gates (OR, AND, NAND, NOT, NOR). Created as teaching material for ADS classes at Faculdade Anhanguera Sobral (April 2024).
Web app for digital product-sample requests, with product image preview, form validation, and a clean two-panel layout (image + form). Dockerized for deployment.
An AI chat assistant for industrial PPCP (production planning & control). Connects to SQL Server and PostgreSQL so operators can query production data in natural language via the OpenAI API, with JWT auth and user management.
Accreditation/credentialing REST API in C# with a clean architecture — Domain, Application, Infrastructure, and WebApi layers fully separated. Dockerized.
Technical mentorship for analysts and a data-driven culture at Grendene.
Documentation and continuous alignment in Scrum/Kanban squads, bridging tech and business.
Query optimization with up to 80% less processing time; batch → incremental pipelines.
End-to-end work — from the frontend to the backend and data engineering.
10+ years keeping up with new stacks (React 19, Spark, dbt) through constant practice.
Shipped Grendene's improvement program, turning data into decisions and real gains.
Business degree + People Management postgrad; turns data (KPIs) into decisions.
Software Engineer and Data Engineer with 10+ years of experience. I work at Grendene S/A — one of Brazil's largest footwear makers — end-to-end: shipping high-performance web apps, designing data pipelines that process millions of records a day, and architecting a cloud-agnostic Data Lakehouse with Stage → Bronze → Silver → Gold layers, data contracts, and automated quality gates.
I'm comfortable across the whole stack: modern React (React Router, TanStack Query, Zustand, shadcn/ui) on the front, NestJS / FastAPI / .NET APIs on the back, and Apache Spark, Iceberg, Trino, Airflow and dbt under the data platform. I also run the infrastructure it all lives on — a self-hosted Docker Swarm cluster with Traefik, Portainer and a local AI stack (Ollama + Open WebUI, RAG over pgvector) serving real factory operations.
