d. Practice - LLM / AI Engineering

LLMs are all the rage. Everybody talks about them, how they will change the industry, even the world. Yet few of us get to work with them in practice on a day-to-day basis. The goal of this workshop is to get your hands dirty with some AI Engineering. We'll take a simple idea and continue adding features to it to force an escalation in complexity until the time runs out.

Duration: ~3 hours
Location: Schraubenfabrik, 1020 Wien
Capacity: 8 people
Time: 13.03.2025 - 17:00 to 20:00

What We Will Do

We will build an application that most of you can likely relate to: A conversational bot that can track grocery purchases for you. While this might sound trivial right off the bat, along the way, we get to touch on and experience a lot of important topics.

• Connecting to OpenAI
• Designing prompts
• Deliberate whether we can run our dev setup locally
• Unit and Integration testing with probabilistic models
• Allowing the model to call tools
• Work with multi-modality
• Get our feet wet with RAG
• Compose multiple models together
• Give our system access to the web
• ... whatever else you can get to.

At the end of the evening, we should be able to leave the hype machine running in the background and have a much more solid grasp on what the challenges involved in building applications delegating and augmenting core logic to LLMs are.

Prerequisites

This is a workshop intended for anybody who is comfortable coding. We will be using LangChain with OpenAI's API. The example code will be provided using Spring Boot. The fundamental issues and abstractions don't change whether you use TypeScript, Python, Golang, Kotlin, or Java, and I don't have strong opinions.

I will be able to help you should you get stuck. I'll be most effective in helping you if you use the JVM as I'm most familiar with it.

With that being said, you will need the following:

• A Laptop
• SDKMan for Kotlin/Java or your stack of choice working and operational
• An OpenAI API Account with a few credits (5$ is more than enough) - https://openai.com/api/

Rough Timetable

Introduction: 0:15 h
Coding: 2:30 h
Sharing: 0:15 h
🍷&🍺:  🤷

Expectations

This is an event amongst peers sharing knowledge to build and nourish a community. Let's make it great together. A bottle of wine or a couple of beers is a good start as a contribution. 😃