AI Wednesday 04 : Introduction to Edge AI
By Sebin Thomas · January 28, 2026
Overview
This week's AI Wednesday introduced Edge AI — running machine learning directly on devices at the edge rather than in the cloud. We covered what makes edge deployment different, walked through a practical implementation on the ESP32, and discussed how quantization helps fit models onto constrained hardware.
Topics
- What Edge AI is and why it matters for latency, privacy, and offline use
- Constraints of edge devices: memory, compute, and power
- A practical ESP32 implementation demo
- Model quantization — reducing precision to shrink models and speed up inference
- Trade-offs between model size, accuracy, and inference speed on embedded hardware
Photos

Highlights
- Seeing a model run on an ESP32 made the gap between cloud-scale AI and on-device inference much more tangible — quantization is often the key to bridging that gap.
Next Week
- Topic: Exploring Mixture-of-Experts (MoE)
- Host: Sebin Thomas