AI Engineer
← All organizations

Database and AI retrieval platform

MongoDB

MongoDB builds a document database and multicloud data platform for developers creating operational and AI applications. MongoDB Atlas combines operational data, search, real-time analytics and AI retrieval. Its automated embeddings, powered by Voyage AI, embed documents as they are written and update embeddings when documents change, helping applications retrieve current context. The Atlas Embedding and Reranking API also makes these models available to applications running outside MongoDB.

Dwight Merriman, Eliot Horowitz and Kevin Ryan founded the company as 10gen in 2007; the database launched in 2009. Chirantan “CJ” Desai is its current President and CEO. MongoDB’s flexible document model is complemented by distributed, multi-document ACID transactions, supported since 2019. Its engineering research uses TLA+ specifications to check transaction isolation and interactions with the WiredTiger storage engine, and measures how much concurrency a transaction protocol permits.

MongoDB serves organizations across industries and more than 100 countries. The company reported over 67,700 customers as of April 30, 2026, including over 66,400 Atlas customers; these counts exclude free users and group affiliated entities together. Atlas supports consumption with minimal commitment, allowing customers to add workloads as their needs grow.

www.mongodb.com ↗

6 talks

Newest first

5 speakers at AIE

Affiliations reflect their AIE appearances, not necessarily current employment.

Apoorva JoshiStaff AI Developer AdvocateRichmond AlakeDirector of AI Developer ExperienceBen FlastMongoDB director of product managementBen PerlmutterAI engineerFrank LiuVoyage AI research manager

Start here

  1. RAG and the MongoDB Document Model

    Start with Ben Flast’s talk to learn how the $vectorSearch aggregation stage, similarity settings, and numCandidates tuning fit into RAG, alongside semantic caching with LangChain.

    Ben FlastAI Engineer World's Fair 2024

  2. Architecting Agent Memory: Principles, Patterns, and Best Practices

    Use Richmond Alake’s talk to understand how persistent storage and retrieval support agentic RAG beyond a chatbot’s immediate context.

    Richmond AlakeAI Engineer World's Fair 2025

  3. AI System Design: From Idea to Production

    Choose Apoorva Joshi’s talk for a repeatable path from initial requirements to production, including deciding where retrieval-augmented generation fits.

    Apoorva JoshiAI Engineer World's Fair 2026

  4. The State of AI-Powered Search and Retrieval

    Frank Liu of MongoDB and the former Voyage AI team explains how semantic, embedding-based search moves beyond lexical retrieval and improves grounding in retrieval-augmented generation.

    Frank LiuAI Engineer World's Fair 2025

Messages from the stage

Retrieval mechanics and quality

Ben Flast explains document storage, HNSW indexes, and vector-search tuning. Frank Liu adds embedding quality, reranking, and application-specific evaluation, connecting retrieval infrastructure to the relevance of returned context.

Memory and perception in agents

Richmond Alake distinguishes conversational, entity, working, short-term, and long-term memory. Apoorva Joshi’s multimodal workshop connects text-image inputs and application-executed functions with Voyage AI embeddings and MongoDB vector storage.

Requirements before implementation

Using a hypothetical claims-review application, Apoorva Joshi ties architecture choices to business goals, human-review constraints, guardrails, and evaluation. Production monitoring includes human override rates and review time.

Affiliations reflect each recorded session, not necessarily current employment. The multimodal workshop recording ends as participants begin the hands-on lab.

Company sources · checked 2026-08-27