847: AI Engineering 101 — with Ed Donner

Fine-Tuning

Fine-tuning AI models is crucial for optimizing their performance in specific applications. Ed Donner explains that AI engineers employ techniques like Retrieval Augmented Generation (RAG) to enhance model accuracy by retrieving relevant information from vast data stores 1. This involves using multi-shot prompting and hierarchical RAG to refine the context provided to the model.

Once a model has been selected, the next step for an AI engineer is to then figure out, okay, how are we going to optimize applying this model to the problem at hand?

--- Ed Donner

Jon Krohn highlights the importance of query-conditioned RAG, which involves rewriting user queries to improve the relevance of retrieved documents 2. This approach significantly boosts the model's ability to provide accurate responses.

   

Agentic AI

Agentic AI represents a proactive approach in AI model optimization, allowing systems to act autonomously beyond simple user interactions. Ed Donner describes agentic AI as suitable for complex problems that can be broken into smaller tasks, enabling models to use tools and iterate solutions 3. This approach is exemplified by AI systems that autonomously notify users of changes, such as price drops in airline tickets.

Agentic AI allows you to be proactive as opposed to reactive.

--- Jon Krohn

Jon Krohn adds that reasoning frameworks, a recent development, enhance agentic AI by guiding models through reasoning processes, thus improving their effectiveness 4.