Developer Conference

What does it look like to run a powerful AI model without the cloud? Kevin Griffin from NVIDIA demonstrates hands-on with a DGX workstation, no internet required. A live look at what on-device AI actually takes, and what it unlocks for Oklahoma builders.

Dr. Kevin Griffin is a Senior Developer Technology Engineer at NVIDIA, working at the intersection of artificial intelligence, high-performance computing, scientific visualization, and advanced software systems. He helps researchers and developers accelerate innovation with NVIDIA technologies across AI/ML, cybersecurity-adjacent computing environments, and next-generation visualization platforms.

Kevin is also the Founder and Chief Technology Consultant at KSG Technology Consulting, where he advises organizations on emerging technology strategy and the practical adoption of advanced computing capabilities. He is currently researching and developing a unified AI agent platform that connects enterprise data, tools, and business logic into a cohesive, intelligent, multi-agent ecosystem.

Before NVIDIA, Kevin spent more than 13 years at Lawrence Livermore National Laboratory in scientific visualization research, software development, and leadership roles supporting mission-critical systems. He also served at the National Security Agency and as a Navy Reserve Information Warfare Officer. He holds a Ph.D. in Scientific Visualization, Computer Graphics, and High Performance Computing from UC Davis, plus graduate and undergraduate computer science degrees from the University of Tulsa and the University of Delaware.

Most engineers have heard about AI tools like ChatGPT, Claude, or Cursor. Many are already experimenting with them. So where does Microsoft Copilot actually fit?

In this demo-heavy session, we’ll explore what Copilot can really do across the Microsoft ecosystem, from everyday assistance in Microsoft 365 apps to newer capabilities like Copilot Cowork that can orchestrate multi-step work across Outlook, Teams, Word, Excel, and SharePoint.

You’ll see how Copilot can help with tasks like analyzing technical documents, summarizing long threads, drafting documentation, and coordinating work across multiple tools.

Expect live demos, honest limitations, and practical scenarios. You’ll leave with a clearer understanding of what Copilot can actually do today, where it fits alongside other AI tools engineers already use, and when it’s worth using or when it’s not.

April is a Power Platform Developer Advocate at Microsoft and a former Power Platform MVP, focused on low-code, Power Apps, Power Automate, Power Virtual Agents, and Copilot. She runs the SharePoint Siren blog and YouTube channel, contributes to the Microsoft 365 Patterns and Practices community, and co-leads the Global AI Tulsa chapter.

Shacole Hamlett is the Founder and CEO of PlanBrite, an AI-powered event pre-production platform for freelancers, agencies, and teams managing complex live experiences.

After more than 15 years producing events and brand activations, Shacole built PlanBrite to turn event briefs into ready-to-use budgets, timelines, run-of-show documents, vendor lists, and production plans. At 200OK 2026, she joined the speaker lineup for a parallel breakout session.

AI coding agents can read context, write code, call tools, and act faster than most teams can review. Luke Crouch from Mozilla will show how language-level attacks like prompt injection, jailbreaks, and malicious context can turn useful agents into risky automation.

This practical security session looks at how developers can stop hacks against their AI, from large language model attacks to coding-agent workflows and secure guardrails. Expect real examples, browser-security lessons, and practices teams can apply immediately.

Luke is a Privacy & Security Software Engineer at Mozilla, maker of Firefox. He has spent 25 years in full-stack software engineering and focuses on the intersection of web technology, privacy, security, and AI. He has worked on data breaches, passwords, online tracking, networking, and cryptographic privacy and security.

Luke is also an advisory board member for University of Tulsa and Oklahoma State University technology programs, previously served as VP and board member for Techlahoma, and continues to help organize Techlahoma events.

Modern AI tools can feel opaque, but the systems underneath them are built from ideas developers can understand. Greg Tatum walks through how language models learn patterns, represent meaning, predict tokens, and become useful chat and coding assistants.

This session traces the path from older translation systems through neural networks, embeddings, transformers, GPT-style models, supervised fine-tuning, reinforcement learning from human feedback, and modern agent feedback loops.

Greg is a Senior Staff Software Engineer at Mozilla. He builds developer-facing tools and explains complex browser, JavaScript, and AI systems in practical terms for engineers who want to understand what is happening under the hood.