In this episode, I sit down with Susana Wong, Staff AI Engineer at Typeform, to talk about her path from front-end to machine learning and AI engineering. We dig into what skills matter in AI today, why evals are so different from unit tests, and how prompt engineering is really more about context and system design than magic wording.
We also explore the Typeform AI agent architecture, with supervisor agents, sub-agents, LangChain, MCP tools, and agent-to-agent protocols, plus the messy reality of running agents in production.
🎙️ Chapters & Topics
0:42 - 7:05 - From Front-End to AI Engineering
7:05 - 9:56 - Underexplored AI Bets
9:56 - 14:03 - Core AI Engineering Skills
14:03 - 21:35 - Building and Running Evals
21:35 - 27:01 - Prompt Engineering Today
27:01 - 29:03 - An AI Engineer’s Toolkit
29:03 - 33:12 - Learning ML Fundamentals
33:12 - 38:52 - Typeform AI Agent Architecture
38:52 - 42:50 - Production AI Agent Challenges
42:50 - 47:17 - ML Pipelines for Front-End Developers
47:17 - 52:27 - Autonomous Agents for Development
52:27 - Local LLMs and Privacy
The wait list for my new project 📙 Neural Networks - From Writing if-else Code to Training AI Models is now open: https://www.js-craft.io/nn/
📖 Neural Networks from Scratch - Presale
I'm writing a book about the timeless foundational concepts of neural networks for JavaScript developers. Go from if-else to weights and biases by building tiny AI models from scratch!
📖 Neural Networks from Scratch - Presale
I'm writing a book about the timeless foundational concepts of neural networks for JavaScript developers. Go from if-else to weights and biases by building tiny AI models from scratch!