536: I've Seen a Giraffe
Oct 15, 2026 · 59m
More Technology podcasts · More Science podcasts
Summary
David Such, author of *Embedded AI*, discusses the practical application of machine learning on microcontrollers, emphasizing a project-based approach over pure theory. The conversation covers selecting appropriate hardware, the importance of sensor fusion and data preprocessing, and the trade-offs between deterministic algorithms and neural networks. Such also shares insights on using AI tools for book production and debugging, while highlighting the challenges of reliability in edge AI systems.
Topics discussed
Introduction and guest background
Rapid fire questions on robots and coding
Book writing process and project-based approach
Target audience and book scope
IMUs, quaternions, and edge intelligence
Engineering vs science: practical models
Hot word project and optimization pitfalls
Hobbyist vs production: handling edge cases
Using AI for PCB design and its failures
Automating book indexing with AI and Python
When to use AI vs deterministic algorithms
Battery state of charge: lookup vs ML
Expert systems and the limits of rules
Selecting hardware for ML: frameworks and memory
MicroPython: use cases and limitations
Neural processing units and robot arms
Curriculum design and depth of coverage
Keeping content current and bonus materials
Advice for writing technical books
Safety in AI, robotics, and the cerebellum
Giveaway details and final advice
More from Embedded
Listen ad-free on Castria