Overview
What this challenge is about.
You receive permission to use the public MovieLens 1M dataset plus a derived item-KG (movie -> genre, director, decade) built from Wikidata. Train two recommenders: a matrix-factorization baseline and a KG-augmented model (e.g., KGAT or a simple embedding-concat approach). Evaluate on Recall@10, NDCG@10, and a cold-start slice (movies with under 5 ratings). Report training time and inference latency. Deliver: training notebooks, benchmark report (5 pages) with a clear go/no-go on the KG-layer investment, and a small fairness slice on per-decade coverage.
The Brief
What you'll do, and what you'll demonstrate.
Decide whether a KG-augmented recommender beats a collaborative-filtering baseline enough to justify the engineering investment.
Earning criteria — what you'll demonstrate
- Compare collaborative-filtering and KG-augmented recommenders rigorously
- Evaluate cold-start performance, not just average performance
- Reason about the engineering cost of a KG layer in a recommender stack
- Communicate research results as a business go/no-go
Program Fit
Where this fits in your program.
Sharpens the same skills your degree expects you to demonstrate.
Skills
Skills you'll demonstrate.
Each one shows up on your verified credential.
Careers
Roles this prepares you for.
Real titles. Real skill bridges. Pick the one closest to your trajectory.
ML Researcher
Owning a controlled benchmark between two recommender families and writing the go/no-go is exactly the work ML researchers ship for product orgs.
This challenge sharpens
- knowledge-graph-embeddings
- recommender-systems
- benchmarking
Applied AI Scientist
Translating a research idea into a cost-aware business recommendation is the day-to-day of applied AI scientists at consumer-AI startups.
This challenge sharpens
- benchmarking
- kg-augmented-ml
- cold-start-evaluation
Machine Learning Engineer
Building the training and evaluation harness plus the engineering-cost estimate is the MLE skillset that recommender teams hire for.
This challenge sharpens
- pytorch
- recommender-systems
- benchmarking