Resources

April 24, 2025

Scaling Vector Database Operations with MongoDB and Voyage AI

11:30 a.m. SGT The performance and scalability of your AI application depend on efficient vector storage and retrieval. In this webinar, we explore how MongoDB Atlas Vector Search and Voyage AI embeddings optimize these aspects through quantization—a technique that reduces the precision of vector embeddings (e.g., float32 to int8) to decrease storage costs and improve query performance while managing accuracy trade-offs. Vector embeddings are the foundation of AI-driven applications, along with powerful capabilities such as retrieval-augmented generation (RAG), semantic search, and agent-based workflows. However, as data volumes grow, the cost and complexity of storing and querying high-dimensional vectors increase. Join Senior Staff Developer Advocate Anant Srivastava to learn how quantization improves vector search efficiency. We’ll cover practical strategies for converting embeddings to lower-bit representations, balancing performance with accuracy. In a step-by-step tutorial, you'll see how to apply these optimizations using Voyage AI embeddings to reduce both query latency and infrastructure costs. Key Takeaways: How quantization works to dramatically reduce the memory footprint of embeddings How MongoDB Atlas Vector Search integrates automatic quantization to efficiently manage millions of vector embeddings Real-world metrics for retrieval latency, resource utilization, and accuracy across float32, int8, and binary embeddings Combining binary quantization with a rescoring step yields near float32-level accuracy with a fraction of the computational overhead Best practices and tips for balancing speed, cost, and precision—especially at the 1M+ embedding scale essential for RAG, semantic search, and recommendation systems

Save your seat →

AI Database Comparison: MongoDB Atlas vs. Elasticsearch

Large language models (LLMs) are revolutionizing AI application development, from retrieval-augmented generation (RAG) to intelligent agentic systems that dynamically reason, learn, and adapt. The AI ecosystem offers a range of technologies to build these solutions—but one of the most critical elements is the database. In generative AI applications, your database directly impacts response latency, application performance, and output accuracy. This session compares two solutions for vector data storage, indexing, and retrieval: MongoDB Atlas and Elasticsearch. We break down how each handles semantic search, provide performance insights, and walk through best practices for implementing vector search in MongoDB Atlas. Watch MongoDB Staff Developer Advocate Richmond Alake as he guides you through: Common search patterns in AI workloads with concrete examples and detailed performance guidance for MongoDB Atlas Vector Search. Why database architecture matters in generative AI, illustrated through real-world implementation scenarios that showcase developer productivity and application capabilities. How to implement MongoDB Atlas Vector Search step-by-step, with live coding demonstrations that show how to enable semantic search for powerful RAG solutions through a unified developer experience. Practical walkthrough of building complete RAG pipelines that integrate real-time data, with working code examples you can adapt for your own dynamic LLM-driven applications. Actionable best practices for optimizing MongoDB Atlas for AI workloads, including live demonstrations of index configuration, efficient query patterns, and practical scaling strategies. Whether you're building AI chatbots, recommendation engines, or agentic systems, understanding how your database powers LLM-enabled applications is essential. Watch now to gain practical knowledge and working code examples that will help you design and optimize AI-powered solutions with MongoDB Atlas.

Watch presentation →

April 29, 2025

AI Database Comparison: MongoDB vs. PostgreSQL and Pgvector

10 A.M. BST Large language models (LLMs) are rapidly reshaping how we build AI solutions. The database technology you choose to power these applications can significantly impact the performance, scalability, and success of your AI application. In this webinar, Staff Developer Advocate Richmond Alake will compare two vector search solutions—PostgreSQL with pgvector and MongoDB Atlas Vector Search—and guide you through selecting the right option for your AI workloads. Whether you’re a data engineer, AI architect, or developer, you’ll understand how to think about and optimize critical metrics like latency and throughput to meet the demands of modern AI applications. What you’ll learn: How retrieval-augmented generation (RAG) boosts LLM-based applications by integrating external data in real time. How PostgreSQL/pgvector and MongoDB Atlas handle high-performance vector operations for tasks like semantic search and recommendation engines. How robust vector databases enable AI agents to reason, plan, and act autonomously, creating truly dynamic and interactive AI experiences. How a real-world application using a financial Q&A dataset illustrates practical deployment and optimization strategies. How key metrics like latency and throughput directly affect the success of LLM applications. Reserve your seat now to be part of this engaging and thought-provoking discussion. Even if you can’t attend, register anyway and we’ll send you a link to watch on-demand and at your convenience. Hope to see you there!

Save your seat →