Thought Leaders in Artificial Intelligence: Edge Impulse CEO Zach Shelby (Part 3)

Sramana Mitra: You have 12,000 developers currently building ML applications, what is the constitution of these developers? Are these 12,000 developers sitting inside larger companies or entrepreneurs building new apps for new ISVs?

Zach Shelby: We have the whole spectrum. That is something that we have strived for. When you bring in new technology in the space, machine learning is advanced for a lot of these industrial companies and developers, we have to be available at the education level. We do have people who are just beginning and learning about this. They are typically professionals, but they might not be applying machine learning in their work. They might be C or C++ programmers but not yet using machine learning.

We recently launched a free Coursera to give those beginning developers some nice course content to learn about what this is. That is the beginning of the community. We have a lot of developers working for startups. We see a lot of startup activity. A lot of startups are building machine learning into real products with the use of our tools. We encourage that because it is free to use.

We also have a lot of larger enterprise customers. We see 4,000 or so companies behind developers in our CRM where we can see who people work for. It’s a big base of larger industrial companies that use these kinds of tools. It’s the entire spectrum in our case. 

Sramana Mitra: Our interest is in the startup community. What is the size of the startup community out of the 12,000 developers?

Zach Shelby: If I had to guess, it would have to be a third of those developers. I would say that at least several thousands are working for startups. Startups tend to be early adopters of technology. They are not scared to go start using machine learning. 

Sramana Mitra: Machine learning, AI, and startups are so hot right now. Tell me about the nature of the application that you are seeing? Can you categorize and segment them?

Zach Shelby: We did a real data analysis last year to look at the top opportunities and top customer engagements that we had. There were three areas that we saw the most activity with machine learning in the edge space. First, we saw it in predictive maintenance in things like industrial equipment on white goods and the electric grid. It’s predicting how a machine is doing. Is it running normally? Does it need maintenance? Has it been broken? It’s determining the answers to these questions and writing them on the equipment.

The second area is asset tracking. It involves tracking mobile things like packages, vaccine distribution, and tracking fragile goods and equipment. This oftentimes involves analyzing the vibration or the movement of the asset and how it is being used. For example, it analyzes how your forklifts are being leveraged in your warehouses. Is it being used as it should be? Are they being used enough that you should have 12 versus 14?

The third area that we are seeing tons of activity is in wearable devices in health and safety. These are devices that people wear on their fingers and wrists in healthcare recovery, elderly care, and worker’s safety. There are tons of applications that analyze people’s motion whether it’s their whole body or their hands. They even analyze ECG to determine how people are sleeping and if they are getting sick.

We have been involved with all kinds of projects like that. Those are the three major application areas. I am going to give you one concrete customer case. We worked with a European manufacturer of smart grid equipment. They put these insulators on the top of a power pole. Those insulators don’t do much. They just insulate the power line from the power pole, but you can put a little sensor on those power insulators. The sensor can pick up the electric current that is going through the lines of its artifacts.

Traditionally, the sensors are really dumb. They might send a report to the power company once a day or once a week. They are slow-moving in that way. The customer worked with our technology and they were able to put machine learning in those sensors. The sensors with machine learning could now detect lightning strikes, branches hitting power lines, ice build-up, or fire in real-time. It determines what is happening at that power pole.

That can be done with ten-year battery life so you can put these things in all the power poles and understand what is happening in real-time. You are based in the Bay area as well and the fire problems that we have had in California have been immense. This is a huge solution for areas that are prone to fire and ice problems and other kinds of damage to power lines. This something that we have recently put into production. 

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