Thought Leaders in Artificial Intelligence: Edge Impulse CEO Zach Shelby (Part 4)

Sramana Mitra: If I am a developer, what are my choices in terms of picking a machine learning engine or platform? What am I choosing from? You and who else?

Zach Shelby: What kind of developer are you? That is the real question. Rationally, machine learning and AI tooling have been designed for data scientists. Almost all machine learning tools are designed for the data scientist. That is okay if your job is to solve back-end cloud problems with machine learning.

For example, somebody gives you a report on the latest finance trends in some product, and your job as a data scientist is to figure out what the trends and problems are in that report. You will use some data scientist tools like a Python notebook with maybe Amazon SageMaker if you want to use a commercial hosted tool or you might just use a generic Ipython notebook on your machine. That is a world of tooling for traditional machine learning in the cloud.

The area that we work with is specialized. We are working with industrial sensors and it goes right on the edge to compute. There are two solutions. You can build your own pipeline. If you are a developer in that space, you have to piece together the open-source tools yourself. Usually, you will need to build and integrate five to ten tools. You can integrate those on one machine or you can integrate them on the cloud.

This is work that you have to do yourself. There isn’t an out-of-the-box solution that just does it all. Things like data collection are separate from training tool kits like Tensorflow. You also need specialized tools to compress that down to a size that can work on a small device. You then need different tools to help deploy that model into the field. You have to use a bunch of tools that you need to put together.

We have a different model because we have an end-to-end solution. We host all that. We are kind of a one-stop-shop for the entire thing as a service. You can get everything pre-installed and integrated and it works. What we run up against are people who want to do it themselves. These are people who want to build their own infrastructure, put all their tools together, and develop their own pipeline for machine learning on the edge. That is perfectly fine if you want to invest the time and effort into it. That is a lot of work. That is usually what we run up to in terms of competition. 

Sramana Mitra: With the thousands of startups building on your platform, can you give us some examples of interesting ones that are finding a lot of traction right now?

Zach Shelby: Yes, we have some amazing startups that we work with. There’s a couple in the wearable space that I would love to point out. We have Neosensory. They are a company that makes a wristband for the hearing impaired. That is important because there are a lot of levels of hearing impairment and it causes a lot of issues with being able to interact with people and the environment.

They put a small device with audio machine learning and vibration elements. You have vibration things all-around your wrist. They have taken this cool application of machine learning and analyzed the audio around someone. You could configure it for the different kinds of sound problems that you might have.

For example, if you want to detect specific traffic sounds or even specific human speech sounds, you can tune it to detect those sounds. You can set it to vibrate for you when the device hears those things. It complements your hearing without having to have hearing aids. It’s an interesting device. We’ve even opened it up for developers so they can go and do their own machine learning and audio algorithms with Edge Impulse to deploy it down to the Neosensory.

That is a great example of a startup. That is where ML is hot right now. Doing what we are doing in building the ML development platform is crazy stuff. There are just a few companies that do that and focus on that. Where the big money is in the application and solving a real problem. Neosensory is a great example of solving that end problem. They’ve applied machine learning with some interesting hardware and deployed that to the end-user. 

This segment is part 4 in the series : Thought Leaders in Artificial Intelligence: Edge Impulse CEO Zach Shelby
1 2 3 4 5