Update talks index and add TurboPuffer talk documentation

- Added a new entry for the TurboPuffer talk, highlighting its object storage-first vector database architecture and cost-saving benefits.
- Enhanced the talks index with a brief overview of the TurboPuffer talk to improve navigation and accessibility for readers.

Author: jxnl

docs/talks/index.md

@@ -86,6 +86,9 @@ LLMs as "pilots that see the world as strings" fundamentally can't understand nu
 
 #### Search Technologies
 
+**[TurboPuffer: Object Storage-First Vector Database Architecture](turbopuffer-engine.md)** - Simon (TurboPuffer)  
+Object storage-first vector database reduces costs by up to 95% while maintaining performance through intelligent caching. Surprising insight: traditional vector databases become prohibitively expensive at scale, but TurboPuffer's tiered storage approach (object storage → SSD → RAM) makes vector search economically viable for massive datasets. Real-world validation: companies like Notion, Cursor, and Linear have migrated to save millions while improving performance.
+
 **[Lexical Search in RAG Applications](john-lexical-search.md)** - John Berryman  
 Why semantic search struggles with exact matching, product IDs, and specialized terminology. Lexical search provides efficient simultaneous filtering and rich metadata that helps LLMs make better decisions. Recommended approach: use lexical search for filtering, semantic search for understanding meaning.
 

mkdocs.yml

@@ -85,6 +85,7 @@ nav:
   - "Talks":
       - "Overview": "talks/index.md"
       - "Search / Indexing / Chunking":
+          - "TurboPuffer: Object Storage-First Vector Database Architecture (Simon, TurboPuffer)": "talks/turbopuffer-engine.md"
           - "Text Chunking Strategies (Anton, ChromaDB)": "talks/chromadb-anton-chunking.md"
           - "Lexical Search in RAG Applications (John Berryman)": "talks/john-lexical-search.md"
           - "Query Routing for RAG Systems (Anton, ChromaDB)": "talks/query-routing-anton.md"