Beyond Filters: Modern Search with Vectors in Django with Kumar Shivendu
summary
Claims a production-ready vector search pipeline that handles billion-scale search, shows an actual integration with Django via a Django semantic search backend talking HTTP to a vector engine running on port 6333, and demonstrates the HNSW graph index and trade-offs: vectors (example 384-dim) give much higher relevance but require more RAM and CPU, with practical knobs like 32-bit to 8-bit quantization (≈4× RAM reduction) and combining vectors plus keyword filters for edge cases.
AI-written summary · about
chapters (12)
- 0:00Modern search with vectors
- 1:30Traditional Django search
- 2:37Keyword search limits
- 5:50Vectors future of search
- 7:58Vector search applications
- 9:45Scaling with HNSW index
- 11:50Vector database setup
- 13:13Django vector search demo
- 16:02Vector search trade-offs
- 18:10Vector search limits
- 19:34Multimodal vector search
- 21:39Summary and future
related talks
29mA Pythonic semantic search
59mContextual search with vector search: exploring your options with open source tools
44mVector Search Beyond the Hype: Real-World Implementation and Performance Insights
27mUnified AI & Relational Power: Scalable Distributed Vector Search with
34mThe Challenges of Making Vector Search Billion-scale
24mVector Search at Uber
31mAdvanced search for your e-commerce website
01h 09mVector Search: Ask Me Anything!