Superpower Your Semantic Search Using Vector Database - From Model Architecture to Deployment
Qdrant - Vector Database & Search Engine via YouTube
Overview
Learn how to enhance semantic search capabilities using vector databases in this technical talk from MLOps Engineer Nicolas Mauti. Discover how Malt, a freelancer-company marketplace, achieved a 5x speed improvement in their automatic matching system through vector database integration. Explore the complete journey from initial requirements analysis to the successful deployment of Qdrant at scale, covering crucial aspects like retriever ranking architecture, semantic space modeling, pre-filtering and post-filtering techniques, benchmarking, and Kubernetes deployment strategies. Gain practical insights into implementing vector databases for improved matching systems while maintaining optimal latency, drawing from real-world implementation experience at Malt. Master the technical concepts behind modern recommendation systems and NLP models through this comprehensive examination of vector database integration in production environments.
Syllabus
Intro
Topics
Retriever Ranking Architecture
Semantic Space
Model
VectorDB
Prefiltering
Post filtering
Benchmark
Deployment in Kubernetes
Latency
Taught by
Qdrant - Vector Database & Search Engine