Olist Data Lakehouse
End-to-end vendor-neutral data lakehouse: Kafka → Spark → Iceberg ingestion (bronze/silver/gold) with Airflow, Great Expectations, Terraform IaC (AWS) and a Streamlit dashboard.
A selection of my recent work.
End-to-end vendor-neutral data lakehouse: Kafka → Spark → Iceberg ingestion (bronze/silver/gold) with Airflow, Great Expectations, Terraform IaC (AWS) and a Streamlit dashboard.
End-to-end real-time streaming pipeline: CSV simulation → Kafka → 3 Flink 2.0 jobs (revenue aggregation, CEP anomaly detection, global KPIs) → Apache Iceberg on MinIO, with Prometheus, Grafana and RocksDB checkpoints.
End-to-end MLOps pipeline: feature engineering → XGBoost → FastAPI serving, with MLflow tracking, Evidently drift detection, Airflow orchestration, Prometheus metrics and multi-cloud deployment (Terraform AWS/GCP/Azure + Helm K8s).
Multi-LLM agentic framework: a LangGraph pipeline orchestrates 4 specialized agents (Architect, Designer, Developer, Tester) to turn a free-text request into tested, documented code, with SVG asset generation and automatic fix loop.
Family tool to preserve burial concessions: centralized tracking of renewal deadlines, automatic email alerts before expiry, and admin space. Public landing page with a contact form for inbound requests.
The site you are currently browsing: a hand-rolled static site generator in Rust and a WebGPU particle background (wgpu → WASM).