Grey Lovelace
Source Allies, Principal Engineer, Coach, and GenAI Specialist
Des Moines, Iowa, United States
Actions
Grey Lovelace is a Principal Engineer at Source Allies, specializing in generative AI solutions and enterprise architecture. With over a decade of experience building resilient applications across insurance, financial services, and agriculture sectors, he has experienced first hand what works, and what doesn’t stand the test of time
As an early adopter in the generative AI space, Grey focuses on cutting through the hype to deliver practical, business-driving solutions. He brings a pragmatic approach to AI implementation, drawing from his extensive experience designing systems that gracefully handle real-world complexity. Grey's expertise lies in transforming emerging AI capabilities into robust systems that deliver measurable value in production.
Area of Expertise
Topics
Behavior Driven Development In Practice
Join me as we talk about the philosophy around Behavior Driven Development (BDD) as an evolution of Test Driven Development (TDD) and how I and my colleagues are using it on a day to day basis.
We will explore how this philosophy can manifest its self in testing strategies and libraries, the pros in long term maintainability, as well as the cons.
We will have a code-along live in the presentation showing how we would implement something like this in a simple API that has multiple requirements we need to make sure are represented, regardless on implementation details.
- 50 minute presentation preferred.
- Video of this talk available here: https://www.youtube.com/watch?v=pOUk65RtNYM&t=2s
- Some technical knowledge required to follow along with the coding portion, but not much. Beginners in python can follow along easily, and if you have experience with any testing or web frameworks in any other languages, you will have no problem.
Measuring GenAI Solutions
Over 85% of GenAI related POCs never make it to production. The most cited reason for this is lack trust in the solution. In this session we explore tools to address this issue, LLM evaluations!
We will show both the philosophy behind how LLM evaluations work, what the metrics mean, as well and how you can use this data to overcome roadblocks to production, and ultimately, providing business value with your solution.
We will cover:
- What does measuring LLM solutions look like?
- Business process
- Timeline aggregation of data to address business problems
- Anecdotal vs objective evidence
- 30 or hour long session are both available
- 5 Minute lightning talk recording of this available here: https://youtu.be/Tda76JqGaNY?si=e6v9Vd02NMZcdO8l
- Attendees will need some GenAI familiarity with concepts like RAG and what LLMs are, but not much beyond that. Business focused attendees often find the information helpful, but the technical side a little too in depth for them.
Making your GenAI RAG Solution Better
So you have implemented a RAG solution, and maybe you even put up evaluations around it. Great! But once you have those things in place, what if your evaluations tell you your solution is not cutting it? Where are the knobs you can turn and levers you can pull in your AI application that are going to make the difference?
Join me as we walk through the 3 main areas changes can be made in a RAG solution to change your outcomes.
- Document Ingestion
- Document Retrieval
- Inference
Talk focuses on the Document Retrieval stage and highlights 3 techniques that can be useful, as well as when you may want to use them:
- Hybrid Search
- Reranking
- Self-Querying Retrievers
Live code examples establishing a baseline, as well as examples of all 3 of these techniques both independently and combined are presented. These examples show how they are put together, as well as results leveraging an open dataset of movie synopses.
- 50 minute session or longer preferred
- Live code examples are easy to follow, but require some familiarity with RAG or python or both.
- Uses this github repo as both a template for the talk, and something people can take home and try for themselves: https://github.com/grey-lovelace/ai-iowa-make-rag-better
Using GenAI to Turn Unstructured Data into Structured Data
Do you have unstructured data sitting around that you can't get value out of? Let's turn it into something useful!
Learn how to navigate pesky mixed-media pdfs and use few shot prompting with images to improve data accuracy.
Follow along with different use cases and live code examples showing each technique!
Outline:
- Introduce use cases and problematic data
- Text parsing vs images and multi-modal models
- Using few shot prompting to address problematic graphs
- How does this scale?
Will need some familiarity with GenAI, but not much. We will cover some basics in the session.
KCDC 2025 Sessionize Event
Open Source North
Measuring GenAI Solutions
Kansas City IT Symposium
AI in Action: How Smart Companies Are Winning (While Others Get Stuck)
Iowa Code Camp Fall 2024 Sessionize Event
Please note that Sessionize is not responsible for the accuracy or validity of the data provided by speakers. If you suspect this profile to be fake or spam, please let us know.
Jump to top