Presented by:

E69b638d643ac997ba6a44d4fe2ad113

Gunjan Juyal

from Google
A41b1dddec6ae9ff674884a3a9981cd9

Saurabh Gupta

from Google
No video of the event yet, sorry!

Using pgvector to build innovative ANN vector-search applications can be an overwhelming experience due to the complexities involved. This training explores techniques to productionize pgvector workloads and build GenAI applications using PostgreSQL, and provides hands-on experience with the most common techniques.

We will cover the following key areas: Estimating the optimal server configuration for your data to be indexed for vector search Key vector search metrics, including the “Throughput vs Recall” curve to balance cost vs performance Create and maintain your vector indexes to get stable recall/throughput performance Create and analyze vector queries, and measure performance impact of data drift Advanced techniques such as filtering and quantization to achieve the desired point in the “Recall vs Performance” curve

Date:
Duration:
3 h 30 min
Room:
Conference:
PGConf India, 2025
Language:
Track:
Training
Difficulty: