scripod.com
Al Engineering 101 with Chip Huyen (Nvidia, Stanford, Netflix)

Overview

Shownote

Highlights

Transcript

Chapters

Pins

Al Engineering 101 with Chip Huyen (Nvidia, Stanford, Netflix)

Lenny's Podcast: Product | Career | Growth

Oct 23
Lenny's Podcast: Product | Career | Growth

Lenny's Podcast: Product | Career | Growth

Oct 23
OverviewShownoteHighlightsTranscriptChaptersPins

Shownote

Chip Huyen is a core developer on Nvidia’s Nemo platform, a former AI researcher at Netflix, and taught machine learning at Stanford. She’s a two-time founder and the author of two widely read books on AI, including AI Engineering, which has been the most-...

Highlights

In this insightful conversation, Chip Huyen shares her deep expertise from years of building real-world AI systems, offering a clear-eyed perspective on what actually drives success in AI product development. Drawing from her experience at Nvidia, Netflix, and as an educator and founder, she cuts through the hype to focus on practical, scalable strategies that organizations can use to build effective AI applications.
08:52
Language modeling is based on predicting the next token using statistical patterns from text.
14:30
Post-training is where the difference lies now
19:57
Data labeling companies face structural economic challenges despite high AR.
29:30
Evaluating each step of information-gathering improves AI summary quality
34:24
Data preparation matters more than database choice in RAG systems.
41:31
Many companies invest in AI literacy but see little employee tool usage
43:21
VPs prefer AI coding agents for business metric gains, while managers prefer hiring more engineers.
48:02
Highest-performing engineers benefited most from using AI tools
52:12
CS is about system thinking, not just coding.
55:31
AI engineers use existing models to build products, lowering the entry barrier and increasing possibilities for AI applications
1:02:16
Voice-to-voice AI models are extremely difficult to build due to latency and naturalness challenges.
1:05:48
Spending more compute on inference can improve model performance by enabling longer reasoning or multiple answer generation.
1:08:23
Look at your daily frustrations as a source for great ideas.

Chapters

Introduction to Chip Huyen
00:00
Chip’s viral LinkedIn post
04:28
Understanding AI training: pre-training vs. post-training
07:05
Language modeling explained
08:50
The importance of post-training
13:55
Reinforcement learning and human feedback
15:20
The importance of evals in AI development
22:23
Retrieval augmented generation (RAG) explained
31:55
Challenges in AI tool adoption
38:50
Challenges in measuring productivity
43:19
The three-bucket test
45:20
The future of engineering roles
49:10
ML Engineers vs. AI engineers
55:31
Looking forward: the impact of AI
57:12
Model capabilities vs. perceived performance
1:05:48
Lightning round and final thoughts
1:08:23

Transcript

Chip Huyen: Our question that gets asked a lot and a lot is how do we keep up to date with the latest AI news? Why do you keep up to date with the latest AI news? If you talk to the users and understand what they want, what they don't want, look into the f...