A preventive health-tech platform I engineered end to end for a Spain-based team. Frontend, backend, cloud infrastructure, and data layer, taken all the way to production.
Production-ready hybrid recommendation service, FastAPI serving layer, modular training pipelines, Redis caching, API-key auth, and CI-ready tests.
Streaming medallion-architecture pipeline: Kafka orders → PySpark Structured Streaming → bronze/silver/gold layers → star-schema warehouse, orchestrated with Airflow.
Production-style RAG backend with tenant-isolated retrieval, local multilingual embeddings, and grounded generation, Django + PostgreSQL 18.
Autoencoder that compresses sensor data on microcontrollers (TF Lite Micro), with a Flask decoder API on AWS EC2 + S3 handling encoded payloads.