How Lettria Unlocked 20% Accuracy Gains with Qdrant and Neo4J

Session Track: App Dev

Session Time: 04:30 pm - 04:45 pm November 6

Session description

In this session, the speaker will present the engineering architecture behind Lettria's GraphRAG system, which delivers over a 20% accuracy improvement while maintaining sub-200ms latency at scale. You will explore the challenges of synchronizing transactional and non-transactional databases, with a focus on designing for high concurrency and building scalable, cost-efficient infrastructure. The talk will delve into the integration of Qdrant and Neo4j, and introduce a custom sync engine purpose-built for real-time data consistency. As a developer, you will gain practical insights into database synchronization strategies, system design for low-latency applications, and efficient use of modern vector and graph technologies in production environments.