The next chapter of agentic infrastructure is memory, persistence, and self-evolving systems.

After bringing builders together at .local NYC, we’re bringing the hackathon to San Francisco during SF Tech Week, with the MongoDB team and our partners, to tackle one of the hardest problems in agentic AI: building harnesses that can adapt, remember, and operate reliably in production. Agents stall in production without governance, memory, and security, so this is your chance to build the infrastructure that gets past that point.

We’re looking for impactful projects built with MongoDB Atlas, including Atlas Agent Engine, our new platform for building, deploying, and governing agents in production. It gives you one foundation for execution, memory, and governance. It also provides long-term memory (semantic, episodic, taxonomic, and procedural), retrieval powered by Voyage AI.

Whether you’re building agents that adapt their own guardrails, systems that hold context across weeks-long tasks, or entirely new approaches to persistent agent memory, bring something that goes beyond the demo and gets to deployment.

🥇 Prizes
$18,000 in cash prizes will be awarded across the three final teams, and more prizes will be announced!

📝 Judges
Jackson Ferrigno, Raindrop

Bosco Albuquerque, AWS

Steffan Mejia, MongoDB

Aram S., Modelence

Note: This event is fully in-person in San Francisco as part of SF Tech Week. Space is limited. Maximum team size is four. All attendees must submit an application and be approved to attend.

The Harness Engineering & Model Wrangling #SFTechWeek Hackathon