Let’s Practice GenAI System Design

There you are, sipping an Old Fashioned and refactoring your codebase when a wild-eyed manager knocks on your door and begs your help to implement a RAG app demanded by The Board. When they call, will you be ready to design a solution with GenAI?

We weren’t. We tuned into podcasts, watched YouTube videos, ran tutorials, tracked https://aigrant.org/ recipients, read research papers, joined Discord servers, debated pipedreams with strangers, and built toy projects until we felt qualified to write GenAI services. All of that theory and technical practice gave us the coveted “how,” but not the “where,” “why,” or “when” of applying GenAI.

To get that good good intuition, we had to write domain-specific business problems and then design solutions that leverage GenAI. The practice worked! It brought confidence to sales calls in domains that we have never worked in and, unexpectedly, made us better at collaborating with each other.

Join us for an interactive solution-design session! We took GenAI use-cases that we built ourselves or saw in the real-world and adapted them into a set of katas.

We’ll work in three stages:
1. Preparation: Form a team and pick a kata
2. Design Phase: Collaborate a solution with your team
3. Peer Review Phase: Discuss the tradeoffs of your approach with the group at large

Learning Goals: 
Participants will...
1. Break a new and complex problem into bite-sized pieces
2. Practice being vulnerable, listening, and communicating novel ideas
3. Discover that a variety of useful and commercially licensed models (e.g., LLMs, embedding models, keyword extractors, sentiment scorers) can be wrapped in containers and used like any other service in a software solution

Target Audience:
- Developers and programmers interested in AI and machine learning
- Tech professionals looking to expand their knowledge of AI tools
- Anyone with a curiosity for AI and its potential impact on the world

Hosts: Gwendolyn Faraday & Lee Harrold