Our universe is made of multiple smaller universes, and when we scale the lens further down to the human scale, we see millions of smaller universes that we want to call tinyVerse. TinyVerse is our daily intersection of physical spaces, devices we co-inhabit with, and the silicon bits that power them.
TinyVerse exhibits some interesting phenomena that we have come to appreciate:
It is more homogeneous than other macro observable universes
Things behave in a more predictable way within a tinyVerse
It can emerge by introducing devices and software in interesting ways in a larger universe
We think tinyVerse can be observed, and with the right filter also measured! You’re probably scratching your head and going this sounds crazy! It’s really not! If you think about it, it is observable and measurable, then it is conceivable because in the tinyVerse some phenomena are also predictable!
Basically, if we have the right model, the right observed data, and at the right place, we can predict the future. Hence, the future is Tiny.
Some examples of tinyVerse around us
We see this all day, every day! Don’t believe us? Well snap, let us tell you!Â
Where you sleep in your own comfy bedroom! Even where you work (Hi There, Pandemic!)
There are devices around you in the room intersecting your daily life. What is that, you may ask? Your apple watch on your wrist! Your Amazon echo sitting on your desk, and your-not-so ergonomics office chair on which you spend 12 hours each day hunching over.
The office where you work, and the naughty kitchen counter with the naughty snacks. Yeah, we will be happy eating a bag of chips or drinking a pop in this tinyVerse. Hell, why not, take both!Â
The local corner store around your home, or that cafe around the bend where you pick up your daily cup of joe. It is tinyVerse with interesting air quality characteristics.
Future is Tiny
Pete Warden wrote a fantastic piece a few years ago as to why the future of machine learning is tiny. His work with the Tensorflow team and tinyML community has been inspirational to us all to start HOTG where we make every edge smarter.Â
We think the future is tiny for several reasons:
Sensors generate more data than we capture or use
5G will bring low latency and high bandwidth leading to higher density of edge devices
Computations will move to the edge, and federate back intelligence
Billions of edge devices to be launched over next few years
(ref: https://www.embedded.com/the-shape-of-the-mcu-market/ Picture courtesy of IC Insights, The 2015 McClean Report)
Introducing HOTG (Hammer of the gods - https://hotg.ai/)
We are excited to tell the world about HOTG - our startup that is trying to make every edge smarter.Â
Hammer of the Gods (HOTG) makes it possible to develop, deploy, and manage apps and models (tinyML apps) on constrained devices. In the next few years, we are expecting a quadratic growth in sensors (especially Arm/Nvidia based devices) and data (5G) resulting in the need for a massive number of developers. HOTG is building the developer tools and the infrastructure layer to bring developers to the intersection of AI, hardware, and tinyML software.
Making things easy, by doing the hard things
We are going to run a container that can run a ML model, and eventually train on real time data, all of which is orchestrated in a distributed way without physical connections creating a virtual mesh guaranteeing resiliency, completely running on an extremely constrained device environment.
Creating a new standard
We are building a new standard for tinyVerse - orchestration for tinyML applications and running on constrained devices. This is going to be the new standards that will be driven and adopted by developers. Engineers, Product Managers, and companies alike will use the tools to make every edge smarter. And best of all? This will be all Open Source.
In order to become a standard, we need to be trusted because we want our trust to be earned, not given. Open Source software has long been the gold standard of how trust is earned. We have to build it. How exactly? We are going to build a vibrant community of developers, early adopters, and leveraging other open communities such as tinyML, rust language, kubernetes, docker, and more! Our aim is to be inclusive of all the tiny waves happening around us!
Here is a developer introÂ
We are tinyML sponsor
Catch us live all week long at the tinyML Summit & Symposium from March 22 to March 26.
Join our Co-Founder, Kartik Thakore in the Partner Sessions on Tuesday, March 23rd from 12-1PM where he will be discussing exciting open source tools, bringing the concept of containerization to the tinyML world. We will have a hangout room shortly after for those who want to participate so hit us up and let's have some fun!Â
Our team will also be at the tinyML Symposium on Friday, March 26th from 10:15AM - 11:25AM to present our research paper on Virtual Machine/Containerized Approach for Scaling tinyML Applications. We will be introducing tools that you do not want to miss out on!Â
Follow us on twitter where we will be tweeting fun tidbits and answering your questions!
Social:Â