YSDA Recommender Systems Course

This repository contains materials for the Recommender Systems course taught at the Yandex School of Data Analysis. This branch corresponds to the ongoing 2025 course.

Syllabus

• Week 1: Intro
  • Lecture: Course overview and organizational details, intro to Recommender Systems problem
  • Seminar: Basic recommenders, user-item latent space
• Week 2: Metrics
  • Lecture: RecSys quality metrics, discovery aspects and evaluation
  • Seminar: Feature engineering, CTR prediction
• Week 3: Candidate generation
  • Lecture: Candidate generation stage, classic RecSys models, ANN-indexes
  • Seminar: CG abstractions, matrix factorization algorithms
• Week 4: Ranking
  • Lecture: Ranking stage and loss functions, diversity and discovery control, feedback loop
  • Seminar: CTR prediction -> LTR, subsampling & reweighting, diversity
• Week 5: DLRM and neural ranking
  • Lecture: Deep Learning in RecSys, ranking models and approaches
  • Seminar: Neural ranking, Multisize Unified Embeddings, Piecewise Linear Encoding, DCNv2
• Week 6: Neural candidate generation
  • Lecture: Two-tower models, sampled softmax and LogQ correction (with derivation)
  • Seminar: SASRec implementation with different losses
• Week 7: Case studies & production
  • Lecture: Real recommender systems design in Yandex and elsewhere
  • Bonus lecture: Graph methods in RecSys
• Week 8: Industry trends
  • Lecture: Generative approach, Semantic IDs, LLMs in RecSys and more
  • Seminar: Best contest solutions presentation

Staff

• Daniil Tkachenko - course admin, lectures
• Roma Nigmatullin - seminars, homeworks
• Kirill Khrylchenko - lectures
• Alena Zaytseva - lectures
• Alexey Krasilnikov - seminars, homeworks
• Artem Matveev - seminars, homeworks
• Vladimir Baikalov - seminars, homeworks