Juan Nicolas Piedrahita Salas
Backend & Data Engineer
Google Cloud Certified Professional Data Engineer. Semi-Senior backend and data engineer in Medellín, Colombia (UTC-5), building production Python systems, data pipelines, and process automation (RPA) across AWS and GCP. Production work includes Apache Airflow platforms on AWS and FastAPI backends that handle frontend-originated user data and inference with Pytorch and ML models.
Selected projects
1 / 6
- Airflow · AWS · S3
Airflow ETL platform on AWS
22 Airflow DAGs on EC2 turn PostgreSQL and MongoDB extracts into Parquet on S3 — about 576 runs a day.
- YOLO · FastAPI · PostgreSQL · MongoDB
Computer-vision speed detection API
A YOLO FastAPI service fed by 9 cameras, about 15k requests over 141 days, 99.91% success, and zero 5xx errors.
- Playwright · GCP · Redis
Distributed trading RPA
Freelance for a trading client: Playwright automation on multiple machines with concurrent instances, GCP cloud Run webhooks, and Redis coordination.
- YOLO · Redis
Real-time CCTV vehicle monitoring
RTSP and ffmpeg ingest, 2 YOLO models at about 30 FPS, a Redis queue, and records in S3.
- XGBoost · scikit-learn
Purchase prediction for a gaming platform
Freelance classifiers on a gaming platform where the data is imbalanced, only about 3.5% of users buy: about 59% recall on buyers on a non overfitted model.
- Firebase
Web platform to access the ML model
Freelance productivity-measurement model for a public-entity research contest, kept on the Firebase free tier.