Me and My (AI) Agent

"Me and My (AI) Agent"
Three Builders. Three Agents. One Bar.
The Lineup

Zachariah Sollenberger — UD Computer Science, Class of 2027
LLM-as-a-Judge: How to Set Up Your Agentic Pipelines for Self-Validation
Zachariah is a rising senior at UD who has published internationally on automatic code validation in agentic workflows and was awarded UD's CIS Junior of the Year. He's building new architectural mechanisms to improve the usability of AI models.

Dong Dai — Associate Professor, Computer & Information Sciences, University of Delaware
From Chatbot to Coworker: What It Takes to Make AI Agents Dependable
Dong leads UD's new Initiative for Dependable Agentic Systems (DAS) and is PI of the Data Intelligence Research Lab. He'll walk through what changes when an AI moves from answering questions to taking real steps in a workflow: accessing data, using tools, and producing work others rely on, and what that means in practice for developers and businesses adopting agents.

Connor Callahan — AI/Software Engineer, Acellus Health (San Francisco)
Cogni-Code: How I Built a Harness-Agnostic Agentic Memory System to Think, Work, and Build Like I Do
Connor builds AI integrations and data-restructuring solutions, with a focus on the insights mined through connecting and reshaping data. He previously built graph intelligence tools for corporate M&A teams.

Format (per speaker)

• 5 min: The Problem. What were you actually trying to solve? Be specific.
• 10 min: The Build. Tools, architecture, live demo strongly preferred over slides.
• 5 min: The Honest Part. What broke, what surprised you, what you'd do differently, what it can't do yet.
• 10 min: Q&A

Wednesday, June 24 · 6:00 to 8:00 PM
UD's IT Research CyberInfrastructure, 550 South College Avenue, Suite 100, Newark, DE