TinyML Blog Posts

I thought I would link my Medium article and Elektor interview about tinyML for anyone interested in perusing. I will post links to other interesting articles on the topic of tiny machine learning on this thread in the future, and recommend others to do so too!


These are very nice posts @ms5113. Thanks for sharing.

I like how you touch on real-world applications, teaching, research and layout the general landscape of TinyML.

I’m (slowly, like really, really slowly) setting up a blog just for tinyML.


It’ll mostly be related to some of the projects I’m currently working on, for example https://github.com/ejri/sensor_networks_azure, and going more in details about building similar systems from scratch.

There’s also an effort from some students in Canada and the US to make it easier/ more accessible for undergrads and high school students. Feel free to join the server about this (invite: https://discord.gg/aCqzKZdC). Any help is really appreciated!!


That’s awesome! Are you all planning on doing a workshop?

Couple of us are getting together to develop a workshop program that can be used by anyone to teach tinyML to others, anyone here specifically being folks with experience in TinyML or have taken the course enough to know the material.

Let us know if you are interested in that.


(Live) workshops and classes. Yes, absolutely, we’d really be interested in working on this with you. I’ll reach out to you by email, and/or we can set up a group call amongst everyone.

Please keep me in the loop about the workshop program. As part of my class project I am trying to coordinate with some local community groups and I would love to be able to offer a workshop on tinyML basics to anyone interested. I wouldn’t feel confident creating a program, but I’m confident I could present a pre-prepared one well.

1 Like

More than happy to keep you in the loop @stephenbaileymagic. We are working with Edge Impulse to develop hybrid material that is based on our edX course + their platform to train the models. That way it is easy to deploy the models.