Don't Let the AI Companies Think for You: Choosing the Right Model Without Falling for the Upgrade Trap

If you are struggling to figure out which AI model is best for your use case or wondering if you should upgrade to the latest model, be sure to check out this exciting talk by Samiksha Kale and Will Angel. Agenda: 6:00 - 6:30 PM - Welcome and mingle 6:30- 6:45 PM - Introductions 6:45 - 7:30 PM - Talk 7:30 - 8:00 PM - Wrap up Description: Newer isn't always better - but how do you know? Every model release comes with benchmark scores, numerous blog posts, and pressure to upgrade. This talk cuts through the hype and gets into real model tradeoffs and performance characteristics such as intelligence, cost, and latency, grounded in real-world examples of newer models not always being better. We'll explore how to actually evaluate models for your use case, when to trust the numbers and when not to. Speaker Bios: Samiksha Kale is an AI Engineer at Leidos, building agents for enterprise automation and tackling the challenge of turning scattered organizational data into a unified intelligence layer. She has worked across a variety of areas and industries, spanning production ML systems in geospatial AI, data engineering in consumer policy research, agentic AI in defense technology, and computer vision in biomedical imaging. Outside of work, she can be found fine-tuning transformer models, hacking together browser extensions, or teaching herself whatever corner of the stack she hasn't explored yet. She holds a B.S. in Computer Science from Carnegie Mellon University. Will Angel is an AI Engineer at DroneDeploy, and is the author of Virtual Power: The Future of Energy Flexibility, an organizer for the Data Visualization and Data Engineers DC Meetups, and the executive director at Data Community DC, a 501c3 nonprofit dedicated to data education in the national capital area. In his free time, Will enjoys wildlife photography, gardening, reading, cooking, art, DIY electronics, and traveling. Will is also the creator of www.vibechart.net.