Description:

Outline:

• Intro to vectors, vector search, and vector DBs
• Vector search internals: HNSW Index
• Why search (relevance) is critical for building great RAG applications
• Re-ranking algorithms
• Fusion algorithms

Takeaways:

• Understand vector search internals and its limitations
• Insights into different models and algos used for reranking and fusion

Prerequisites:

• Must have: Basics of python, search, and ML
• Nice to have: Vector search, RAG

Content URLs:

Slides

Speaker Info:

I love tinkering with search, recommendations, and RAG. I have worked at multiple early stage startups like SuperTokens (YC 20) and FamPay (YC 19). I discovered my passion for search during my GSoC internship, where I built advanced search features on the top of metadata from 250M+ open source repos for the Software Heritage organization.

I currently work as a Software Engineer at Qdrant (OSS Vector Search Engine written in Rust 🦀) and I love to give deep technical talks to share my knowledge. I've given multiple talks in the past at meetups like FOSS United, Tensorflow User group, Microsoft Reactor, etc.

Speaker Links:

• https://kshivendu.dev/bio
• https://kshivendu.dev/linkedin
• https://kshivendu.dev/twitter
• https://kshivendu.dev/talks
• My first talk's video (3rd highest viewed talk video on FOSS United Bangalore channel)