```
Welcome back IBM Data AI meetup community. If you have ever watched an AI agent
confidently hallucinate its way through a task, and wondering is there a way out?
Come and learn about open source project MeLLeA (a Python library) to see how it
can help with some of the issues associated with leveraging LLMs to accomplish your task!
========================================
WHAT IS THIS
========================================
Georgia Tech and IBM are co-hosting a hands-on technical session on building reliable,
auditable AI systems using MeLLeA — a Python library from IBM Research Cambridge (mellea.ai/) —
alongside IBM Bob and Granite open source models.
This is a working session. Bring your laptop. There will be hands-on opportunities!
========================================
WHAT YOU WILL SEE
========================================
-> MeLLeA in action — Instruct, Validate, Repair loops that make LLMs predictable
-> A real demo: legal review checker built with MeLLeA to flag AI-misattributed citations
in automated document pipelines (built by Georgia Tech researchers Adrian Lozada and Di Wu)
-> Spec Driven Development — write the spec first, let the AI build to it
-> Leveraging IBM Bob — build your own agentic AI dev partner
-> Open source data + IBM Granite models in the mix
========================================
AGENDA
========================================
12:30pm -> Networking and arrival
1:00pm -> OSPO@GT — Open Source AI Introduction
1:05pm -> Generative Computing: Building Predictable, Auditable AI with MeLLeA and Granite
2:15pm -> Break
2:30pm -> Bob onboarding session
3:15pm -> Break
3:30pm -> Spec Driven Development with Bob and Open Source Data
4:30pm -> Networking and open discussion
Agenda is subject to change — check this page for updates closer to the event.
========================================
WHO SHOULD COME
========================================
Developers and students obviously but anyone interested in or working on