Each project includes the technical story, key decisions, and trade-offs. Click any card to dive in.
A medallion-architecture data lake on AWS S3 with Prefect-orchestrated ETL Lambdas — raw API responses → bronze joins → silver analytics table.
An end-to-end ETL pipeline that ingests movie metadata from IMDb bulk files and a REST API, stages it in MongoDB, and lands it in a normalized relational schema with foreign keys.
Cloud-based WhatsApp chatbot for medical appointment management using LLaMA 3.3-70B for intent parsing, AWS Lambda for serverless orchestration, and Google Calendar as the scheduling backend.
A full BI stack analyzing 8,847 Airbnb listings — from raw CSV load through SQL modeling on Supabase to stakeholder dashboards in Preset.
A deep exploratory analysis across 4,000 movies to identify the factors driving box-office success — framed around a low-budget-production investment scenario.
A distributed ML pipeline processing 17M Amazon reviews with PySpark MLlib on AWS Glue — including S3 medallion storage, feature engineering, and model serialization for batch inference.
A continuous ingestion pipeline for Spanish grid demand with InfluxDB storage and Prophet-based day-ahead forecasting. Includes a dashboard for real vs forecast visualization.
A pandas-driven analysis across 13 relational CSVs (75 years of F1 history) with multi-way joins, filtering, and map-based visualization.
Three ML-driven marketing projects in one notebook — telco churn classification, customer lifetime value regression, and RFM-based customer segmentation.
A two-part project on the Ames Housing dataset covering data prep, Ridge/Random Forest/KNN modeling, SHAP/PDP explainability, PCA/t-SNE, and Gaussian mixture clustering.