The MAD Podcast with Matt Turck What You MUST Know About AI Engineering in 2025 | Chip Huyen, Author of “AI Engineering”
115 snips
Jan 16, 2025 
Chip Huyen, an esteemed AI engineer and author, shares his insights on AI engineering. He distinguishes AI engineering from traditional machine learning, explaining how foundational models democratize app development. Huyen dives into the crucial role of prompt engineering and the complexities of AI model evaluation. With a closer look at the generative AI stack, he dispels myths around Retrieval-Augmented Generation (RAG) and discusses the evolving nature of AI agents. This conversation is a treasure trove for anyone interested in the future of AI technology.
AI Snips
Chapters
Books
Transcript
Episode notes
Model Size and Data
- Larger models, with more parameters, possess greater learning capacity, but require more data to maximize potential.
- Chip Huyen explains that training a large model with a small data set is a waste of its capabilities.
Sampling Techniques
- Sampling helps improve and debug language model applications by influencing output selection.
- By adjusting sampling parameters, developers can nudge models towards more creative or predictable responses.
Evaluation Metrics
- Evaluate AI systems based on clear business metrics tied to ROI, like increased purchase rates or fraud detection success.
- Choose use cases with easily measurable outcomes for easier enterprise adoption.
