Free Lesson
Vector search from zero to production at GitHub
Part of The Frontier of AI Search
60 min
Jun 24, 2026 12:45 PM
Virtual (Zoom)
In this video
What you'll learn
Plan semantic search beyond the MVP
Why early vector-search MVPs can mislead you, and what to test before production.
Debug vectors at production scale
How capacity, indexing, sharding, reindexing, and Lucene merges complicate rollout.
Tune retrieval for real users
How scoring, oversampling, num_candidates, and linear retrievers affect relevance.
Why this topic matters
Vector search looks easy in demos, but production scale exposes hidden failure modes: capacity planning, indexing bugs, quantization tradeoffs, sharding, and ranking quality. GitHub’s BBQ rollout is a useful case study in turning “just add vectors” into an operationally reliable semantic search system.
You'll learn from
Previously at


