Backend & data engineer. I work where the volume starts to hurt: time series, ingestion pipelines and APIs that have to answer fast over real data.
Currently backend and data at Trin Hub, and co-founder at Seu Canto Digital and MedPrintAI.
Time-series backend at scale. Remodelled a data platform onto PostgreSQL 16 + TimescaleDB hypertables and wrote the migration that validated 3.5M rows out of MySQL. Production dashboards went to 11-60ms over 2.9M real rows, with runtime JOINs off the read path entirely.
A 35+ DAG Airflow pipeline. Handles peaks of 50k+ events/day: scraping, Kafka ingestion, persistence, media download, OCR, transcription and LLM analysis.
Diálogos pelo Piauí. Citizen-participation platform behind the state's 2024-2027 budget plan. I modelled the domain and built the Django REST APIs that carried 7,053 sign-ups and 421 elected proposals across 12 territories. 2nd place at Réplica Piauí 2025 and catalogued on the federal government's VitrineGov.
Dados PI. The state's open-data platform and thematic observatories, ~3,600 monthly users. 3rd place at the same award.
MedPrintAI. Turns cranial CT scans into printable 3D models for surgical planning, in seconds. FastAPI for DICOM ingestion, anatomically calibrated mesh cleanup, and PyTorch / nnU-Net v2 for segmentation. Selected for Catalisa ICT (Sebrae).
Most of this lives in company and government repositories, so the code isn't here.
Python FastAPI Django REST SQLAlchemy PostgreSQL TimescaleDB Redis
Apache Airflow Apache Kafka Docker AWS Terraform Keycloak Next.js TypeScript