While at AWS, I worked with enterprises and start-ups to help accelerate their IoT ideas to market. While most customers were building connected solutions, the promise and inherent value of a disconnected (or poorly connected) edge was evident to those of us working in this field. Frankly, other companies (Cisco in particular) has built whole businesses on the edge (fog/mist computing). That said, most companies treat the edge as an extension of the device and keep it separate from the cloud.
As my team was preparing AWS Greengrass for launch and working with early adopters, it became clear that AWS Lambda as a common programming model between the cloud and the edge was a significant paradigm shift. It also became clear that current partner and customer ecosystem was not prepared to take advantage of this shift. This was the original intuition behind TensorIoT.
What is the problem?
IoT devices today generate vast amounts of data and in many cases this data needs to be processed in real time. While Cloud is the right setting for deeper enrichment and analysis, most of the inference and decision making needs to be made on the edge. Designing, developing, testing and deploying applications to millions of edge devices in a reliable, repeatable way is complicated.
In addition, as explained in the introduction, the current architecture at customers separates the edge and cloud IoT implementations. This leads to a complete separation between field and control room technologies. Even in connected systems there is a significant lag between real-time data and actions. The absence of a standard architecture leads to various communication protocols and systems which requires experts in each of these domains to interpret the data/systems.
What does TensorIoT offer?
TensorIoT aims to brings the agility and manageability of cloud to IoT and Edge solutions. In practical terms, we are helping customers build cloud-native, microservices based applications which may be deployed to serverless computing platforms, containers or pushed to the edge using the same development toolchains.
We have also developed CICD pipelines for machine learning models which allow your data scientists to develop, train, test models in the cloud and deploy them to edge devices for inference with a single click. Our solution takes advantage of the unlimited processing power available in the cloud while leveraging the rich deep data sets on the edge. In our architecture, the models are constantly refined and can run on the cloud and on the edge.
What are the benefits for our customers?
We have a proven track record of optimizing production pipelines to increase revenue while decreasing operating costs. Our ML solutions help increase asset lifetime while reducing maintenance costs. In a number of industrial use cases, we have demonstrated enhanced safety and improved environmental record of solutions.
In addition to our IoT and ML experience, we also have a proven record to integrated voice-enabled capabilities into solutions to make them more accessible and useful for end users.
How do customers start?
For customers with IoT use cases, we offer a comprehensive suite of AWS focused open source cookbooks to accelerate their IoT deployments. We offer a range of services from strategy, architecture to development and long-term maintenance of these solutions.
Many of our customers have complex legacy applications, and we have built a rapid ‘Monolith to MicroServices’ migration path to optimize and improve such applications.
As the launch partner for AWS Serverless Application Repository, our team also offers deep expertise in migrating your applications to a serverless computing platform.
What’s with the name and logo?
We firmly believe that Machine Learning on the Edge will be the killer app with immediate business outcomes. The applications of this are much wider than IoT, but we wanted to highlight our unique IoT experience as well. With that focus, the team settled on the name TensorIoT pretty early on.
Our original brief to award-winning UK based designer Waqas Dogar was to merge IoT, Cloud, AI, MicroServices and also highlight the virtuous cycle of using them together.
Below are the high level drafts we went through as we got to the final logo design.
So, that’s the background of the name and logo, which I now realize is probably TMI for most of our readers.