Enhancing Vector Databases with Real-time Stream Processing
summary
Unified realtime data platform Hazelcast combines in‑memory storage and stream compute to augment a vector search demo that uses Quadrant vector DB and a Jupyter notebook: embeddings are loaded, a text query returns hits, then results are filtered by attributes (example vote_count>100) and by many‑to‑many maps of movie→cast→actor to narrow matches. Platform runs on ≥3 nodes, offers near‑zero access latency, embeds Python and ONNX models, and supports a 5‑model majority‑vote fraud pipeline; demos run and produce expected filtered results.
AI-written summary · about
chapters (7)
- 0:00Introduction to talk
- 1:50Hazelcast platform architecture
- 7:08Demos overview
- 12:08Similarity search demo
- 20:23Fraud detection use case
- 22:52Fraud detection code walkthrough
- 29:04Conclusion and Q&A
related talks
27mUnified AI & Relational Power: Scalable Distributed Vector Search with
28mThe Memory Efficient Indexing for Vector Databases
36mUnleashing the Power of Vector Databases for AI-Driven Applications
48mA fun and absurd introduction to Vector Databases
50mA fun and absurd introduction to Vector Databases
35mVector Databases: Using Vector Search with Graph and Key-Value Queries
01h 23mVector Databases Demystified: Discover how they work
29mFrom Quick Wins to Revolutionising Productivity & CX with GenAI