Article series · 2 lessons
Recommendation systems
Follow signals through candidate generation, ranking and evaluation.
Read the series introduction →
Follow the sequence below, or pick up where you left off. Nothing is locked.
What you’ll work toward & your progress
What you’ll work toward
- Candidate recall, exposure bias and constrained ranking
- A ranker has excellent precision on previously shown items. Why might it fail on a newly launched catalogue?
Completion is stored on this device only. Nothing is locked; start where it makes sense.
Series preparation
Explicitly listed by the series introduction; lesson order does not create a prerequisite.
Before the first lesson
These are the starting lesson’s prerequisites, not requirements for every advanced topic below.
How to practise this subject
Work a small ranking example and compare it with a baseline. Explain what an offline metric measures, what it misses, and how the training cut avoids future information.
Lessons in order
- 01
Learning how to build an recommendation system from initial signals
Build candidate and propensity-model evaluation contracts without future leakage. Mechanisms, worked examples, failure analysis and complete practice answers.
- 02
Monolith: a source-backed study of real-time recommendation architecture
Explain full-key embedding admission, expiry and incremental synchronization. Mechanisms, worked examples, failure analysis and complete practice answers.