## Round 1 - ML / System Design Design a movie recommendation system for a streaming platform with 10M users and 100K movies. **Approach 1 - Collaborative Filtering (user-based):** - Build a user-movie rating matrix (sparse). - For user U, find the top-k most similar users by cosine similarity of their rating vectors. - Recommend movies highly rated by similar users that U has not seen. **Approach 2 - Matrix Factorization (ALS/SVD):** - Decompose the sparse matrix into user factors (U, d) and item factors (V, d). - Predicted rating: `r_ui = U_u . V_i` - Train with alternating least squares or SGD. ```python def cosine_similarity(vec_a: list[float], vec_b: list[float]) -> float: dot = sum(a*b for a,b in zip(vec_a, vec_b)) norm_a = sum(a**2 for a in vec_a) ** 0.5 norm_b = sum(b**2 for b in vec_b) ** 0.5 return dot / (norm_a * norm_b) if norm_a and norm_b else 0.0 ``` ## Follow-ups 1. How do you handle the cold start problem for a new user with no ratings? 2. What is the difference between explicit feedback (star ratings) and implicit feedback (watch time)? 3. How do you evaluate your recommender offline before A/B testing? What metrics do you use? 4. How does your system handle a movie released today with no ratings yet?

InterviewDB Question

Movie Recommendation: Design a Collaborative Filtering Recommendation Engine

Question Details

Round 1 - ML / System Design Design a movie recommendation system for a streaming platform with 10M users and 100K movies. Approach 1 - Collaborative Filtering (user-based): Build a user-movie rating matrix (sparse). For user U, find the top-k most similar users by cosine similarity of their rating vectors. Recommend movies highly rated by similar users that U has not seen. Approach 2 - Matrix Factorization (ALS/SVD): Decompose the sparse matrix into user factors (U, d) and item factors (V, d).…

Full Details

🔒

Unlock all Karat questions

Full insider details, leaked discussions, and candidate experiences.

or every company, $100/year →

About This Question

This is a reported interview question from a karat interview.

It covers the following topics: System Design, Coding, Matrix .

Topics

System Design Coding Matrix

Movie Recommendation: Design a Collaborative Filtering Recommendation Engine

Question Details

Full Details

About This Question

Topics

More Karat Interview Questions