Tiny Models, Big Curiosity: The ATHER TinyML Workshop at MACE
We usually picture AI as something massive that runs in the cloud. On September 12, seventy-two students met a much smaller version. It ran on a chip the size of a coin, right there on their desks.
That was the whole idea behind ATHER, a full-day hands-on TinyML workshop held as part of Chrysoderas, the technical conference organized by the ISTE MACE Student Chapter at Mar Athanasius College of Engineering, Kothamangalam. I had the chance to lead it as a Seeed Ranger.
Why TinyML, and why it matters
We started with the big question: what is TinyML, and why should anyone care? The short answer is that TinyML lets machine-learning models run directly on small, low-power devices. There is no cloud in the loop and no constant internet connection.
That single shift changes a lot. Models respond instantly because the data never leaves the device. They keep working offline, in places where connectivity is patchy. They sip power, so they can run for a long time on a small battery. And they keep data private by design. To make it concrete, we looked at real uses of edge AI in areas like environmental monitoring, safety, and accessibility. You could see the ideas starting to land.
From idea to working model
Then came the part everyone was waiting for: building something real. Using Seeed Studio’s SenseCraft AI platform, students went from a blank screen to a working model in minutes, with no heavy coding required. They started with image classification, teaching a model to tell one thing from another. Then they stepped up to object detection, watching the model find and box objects in a live camera feed.
All of it ran on real hardware. The Seeed Studio XIAO ESP32S3 Sense and the Grove Vision AI V2 module became the stars of the day, making sense of the world in real time from a board that fits in your palm. For many participants, this was the first time they had seen a machine-learning model run on something they could hold in one hand.
To finish, we moved into Edge Impulse, the platform for building fully custom TinyML models. This was where it all came together. Every participant collected their own dataset, trained a model on it, and deployed that model onto real hardware. By the end of the day, nobody was just watching a demo. Each of the 72 students had taken a model through the full journey, from raw data to a working file running on a device in their hands.
The room came alive
With 72 participants, the energy was hard to miss. Questions came quickly, and they got sharper as the day went on. Students moved from “how does this work?” to “could I use this for my own project?” in a matter of hours.
That shift is the whole point. By the end of the day, the group was not just following steps. They were thinking like builders, connecting what they had learned to problems they actually cared about.
Wrapping up
A big thank you to the ISTE MACE Student Chapter for hosting ATHER as part of Chrysoderas, and to every student who showed up curious and left inspired. Days like this are a reminder of how quickly interest turns into capability when you put real hardware in people’s hands.
ATHER, a full-day TinyML workshop, was held on September 12, 2026, as part of Chrysoderas by the ISTE MACE Student Chapter at Mar Athanasius College of Engineering, Kothamangalam, with support from Seeed Studio.
About Author
