Granite Meets Mellea: Engineering AI You Can Trust

IBM Research Cambridge and the MIT-IBM Lab present another installation in our “Build-n-Brew” series:

Granite and Mellea: Building Predictable, Auditable AI Systems

In this session we'll begin with an overview of IBM’s Granite model suite – lightweight, performant models, released under an Apache 2.0 license and designed for scalable, enterprise workloads.  Then we’ll introduce Mellea – an open-source library that treats large language models, not as conversational interfaces, but as programmable components inside real applications. You'll see how Mellea is used to implement workflows that are predictable, auditable, and reproducible. Combined with smaller open-weight models like Granite, developers can do “Frontier Model Things” but with better robustness and at a tiny fraction of the cost.

We’ll be rolling up the sleeves.  You can listen and learn, or you can bring your laptop and follow along.  This session is open to developers, engineers, AI builders, and managers who are interested in leveraging the latest in open source LLMs and AI tools to build, not demos, but software that scales.

Doors open at 3:30p, and we’ll get started with the technical “Build” session at 4:00p. At the conclusion, we’ll all head over to one of the local watering holes for an informal “Brew”. 

Registration required.

REGISTER HERE