capstone4–8 weeks~4h mentor timeLive defense
ML Recommendation Engine
Build and evaluate a production recommender with offline metrics and an online serving path.
By DataWicket
pythonpytorchfastapi
The brief
Build and evaluate a production recommender with offline metrics and an online serving path. The brief is deliberately incomplete — you make and justify the tradeoffs.
Deliberately ambiguous — you make and justify the tradeoffs. Your work is checked against a held-out evaluation you can't see in advance.
Milestones & mentor checkpoints
1Data & baselineweek 1mentor review
2Model & evaluationweek 3mentor review
3Serving & write-upweek 6mentor review
How you're graded
Model quality
40% weight
Engineering
30% weight
Analysis
30% weight