Skip to content

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

Group photo

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