Anime Recommender System
A recommendation is only as good as the pipeline delivering its data.
The Challenge
Content recommendation engines face two primary challenges: cold-start content discovery and real-world deployment complexity. Most tutorial systems focus only on offline algorithm scoring in a notebook without addressing how raw data gets extracted, cleaned, loaded into an enterprise warehouse, and served via containerized microservices.
This project was engineered to simulate a full production pipeline: ingesting millions of user ratings, transforming them with Talend ETL into an Oracle Autonomous Data Warehouse, training collaborative filtering models in Python, and serving recommendations from a Dockerized microservice.
Data Pipeline & MLOps
Connecting data engineering with applied machine learning across four discrete stages:
Recorded Demonstration
Screen recording demonstrating database connections, model scoring, and the interactive web recommendation output.
Next Project
Explore computational humor and language model prompt design.
Next project: Fact → Joke Generator →