Mikiko Chandrasekhar
- 1 episode on Chain of Thought
- 7 topics covered
- 1 full transcript
- Jul 2025 first on the show
At the time of the conversation, Mikiko Chandrasekhar was a Staff Developer Advocate at MongoDB, with a background spanning the full data-to-AI pipeline. On Chain of Thought she laid out MongoDB’s vision as the memory store for agents, supporting multi-agent systems through storage, vector search, and the Voyage acquisition.
Episodes
-
EP 33 · AI Agents
Mastering Multi-Agent Systems | MongoDB’s Mikiko Chandrasekhar
What the conversations cover
These notes summarize the episodes featuring Mikiko Chandrasekhar, including other speakers in panel conversations. Follow a source link to read the discussion in context.
EP 33 Mastering Multi-Agent Systems | MongoDB’s Mikiko Chandrasekhar
- At a poll Mikiko Chandrasekhar ran during her AI Engineer World’s Fair lightning talk, roughly 80% of the room said they were building agents and nearly as many had agents in production — but when she asked who had gotten their agents working 100% of the time on the first try, not a single hand went up. That gap is where she locates the reliability problem. Episode takeaways
- Mikiko frames MongoDB’s role as the memory store for agentic systems — not only vector data for semantic search and RAG, but the chat logs and traces developers need to debug errors and keep improving the underlying data and processes. She points to the Voyage acquisition, which brings high-performing embedding and re-ranking models in-house. Episode takeaways
- Treat agents as software products, Mikiko argues, rejecting the let them do their thing and catch the errors later on the application layer school. Non-determinism changes some things, but established software-engineering and MLOps best practices still apply with rigor — she credits Hamel Husain and Shreya Shankar as voices for that discipline. Episode takeaways
In their words
If you don't have a way of observing what these agents are doing, if you don't have a way of evaluating them, and if you don't have a way of putting guardrails around their behavior, you just don't have an agentic system that's gonna really survive.
Mikiko Chandrasekhar Mastering Multi-Agent Systems | MongoDB’s Mikiko Chandrasekhar · 0:00 Transcript excerpts; ellipses mark omitted words. Some automatically selected quotes have filler words removed.