Krishna Rangasayee, CEO & Founder, Sima.ai, Amit Kumar Mitra, Sr Director, SiMa.ai & Sudershan Vuruputoor, Site Lead, SiMa.ai takes us through the different use cases and applications where the solutions are being deployed.
What is the story behind the genesis of SiMa and what are the solutions and services in your portfolio?
SiMa was started in November 2018 with 60+ people and was working mostly in the Bay area and we recently started our site in Bangalore. Our lead investors are Dell Technologies, Amplified partners and link. In the past 10-15 years machine learning has played a very big part in reshaping the cloud. We use Google Maps or Apple Maps to go even the smallest distances, so you can understand how cloud has changed and effected our lives.
In the next 10-15 years, the larger market will be Edge and machine learning will play a very big part in re-shaping everything in the Edge. Our primary focus is to build a purpose build machine learning platform for the Edge. We believe that is the missing link that is preventing the scaling of Edge. That is the intent and vision behind what we have started and it has been 2 years and we are very close to production.
What are the use cases and applications for AI/ML that can be leveraged for scaling up?
We have 3 market priorities –a large category is on smart vision these are security surveillance, health metring, security applications, retail safety. The second one is robotics and drones and the third one is autonomous computing. People have deployed an application for lot of infrastructure and due to the pandemic people have added a new capability added to the existing infrastructure. They want to add thermal monitors, social distance monitors in the security cameras. These require a lot of high performance machine learning and they want to leverage the existing infrastructure by creating new applications with new capabilities. The third category of use cases we see is around public safety and we make sure in high security areas people are identified only from facial recognition.
What will be the business model to be followed for the surveillance part?
There is a well-established chain of integrators that sell into deployment and the infrastructure of how to roll out security and safety structure well in place. We are engaging with the existing supplier and enabling them. But these are new products for them so we are working with them and helping them. I think there will be an overhaul in how people are doing security and surveillance infrastructure happening today.
What is the blend between brownfield and greenfield deployments?
The use cases that were preciously shared require new equipment and an overhaul of both the hardware and software. We are seeing benefits in North America and portions of Asia. We are still gaining customers from North America and portions of Asia, but as the deployment kicks in globally we will learn from what we are doing right now and what we can do different and better. So for me it is a blend of both.
What about the update on the robotics and autonomous computing parts?
Robotics has a broad category of use cases. The area where we see the highest volume is impactful in the automation and warehouse logistics. There are also lot of robotics market where we are seeing a lot of interest in. For factory floor, logistic robots they are using old technology today where the robot is like a car with a path finding technology and human safety run and environment around it. Robots had a work time of 45 minutes and they had to go back to the docking stations and recharge. We are showing customers how that time can be 6-8 hours based on our technology. Not only big companies but lot of start-ups, are using robotics. So because of the volume and scale in the regulatory elements, people are looking to deploy that.
With multiple regulations prevalent across different geographies, how do you ensure compliance?
It is going to be a bigger challenge going forward. As the volumes are there the regulations will happen. The fact that the governments are thinking about regulations is because the high cycle we had in the past 10 years of possibility there is no need to regulate any more. People need drones, robots, better smart vision systems there is no doubt need for regulations. This is not new for us we have lived with the regulatory bodies.
The autonomous industry has done a good job with it and a lot of learning with it and people like Amit and Sudershan have gone through this certification with lot of product in different markets. So I am sure we will learn and it will be more complicated than today.
What are some active use cases you have in different geographies?
Both in America and Europe delivery drone have begun to move scale. Factory floor and logistic robots in the North American market and have beginning to see a good ramp up in Japan and South Korea. We are looking heavily on the European market and have not deployed in the china market so can’t speak about that market as of now. From regulatory perspective we have established with Europe, North America, Japan and South Korea. A lot is happening in the Indian market and the needs of the India market and it is surely different from the other geographies.
What is SiMa’s role in developing workforce with relevant skillsets?
The number of people who understand the scaling of machine learning is a small subset. We are very lucky to identify those talents and grow our team. Attracting talent is the key element and we are able to do that and also understand that North America is not the only geography where talent is.
As we are starting our site in India now we want to find people who know machine learning and have the skillset to groom and grow this. The immediate term we want access to people who not can work in development but can also interact and connect with customers. We already have ten people who can work with the customers as they also need help. We are a young company and we need to focus on solving the problems, not all of them at once but some problems which are solvable. We are taking one step at a time and solving two key challenges one is to get our products into productions and two is to help some of the best customers in the world.
What is the update on the design and development centre and how are you looking at India as a market?
India has a great talent pool in almost everything we do be it hardware design, software design etc. In the last 10 years India has a lot of experience in systems not just in technology but in individual aspects and solving applications. At the end of the day the company is defined by its people the strength of the people we have. We are starting small in India, Amit is heading our software efforts and Sudarshan is heading our hardware efforts. The site in India is going to be for more than development we will be doing systems application and lot of other key things for both the Indian market and other markets there.
It’s a very logical choice to have a site in India, because you name any technology company OEM, technical service providers you have all the ecosystems in Bangalore. The access to the talent pool is something which is very dynamic and when you talk about machine learning and computing there is a huge hype about it but people are also interested to learn and adapt into it very quickly. So the willingness to learn and be a part of it is the best part and it helps us a lot as well.
Does the India centre also provide support and services to the global customers?
That’s the plan one the production starts there will be a business and application support team who will work from here and support the customers in the different time zones. Then understand our technology and help it adapt in other way understand the legacy application and help them out.
The complete AI-ML solution that we are trying to do requires expertise in multiple domain not just the ML & AI that’s just the front end it includes the computer architecture, computer vision etc. they all have to come together to provide a solution. In a survey I found that 10% of all the major international companies like Intel and all are Indians. This is because there is a large talent pool here and this replenished very quickly as there are almost every year nearly 250,000 engineers coming from the different institutions. So this is what every company is looking for we are also utilising that here at SiMa also.
What are the alliances you are looking at with other companies both for design and production?
We have three technology partners one is public adobe, arm technologies and from a computer vision EDA IP tool infrastructure we also have a strong partnership with synopsis. We will be fabricating our products with TSMC. From the technology point we feel very good with these three industry leaders. In terms of leveraging with Intel and others we actually compete with them. So as we are competing with them we are careful with who we partner our technology with. Israel has always been very fascinating for me there are many parallels between Israel and Bangalore. We are absolutely engaged with many market leaders from Israel.
How will be SiMa’s GTM roadmap going to be once production attains critical scale?
As we are developing we are thinking about our customers. One of the market complexity on edge is unlike cloud is that they have thousands of customers in some place so to bring technology to scale is a very different problem than in cloud. The state we are in our goal is to win our first ten customers, there is a lot of business development people but what we need are a lot of technical systems knowledge. Then its goes to 50 where we need more of a marketing infrastructure and business development infrastructure. One it goes past 100 or 1000 we need to figure out a relationship with representatives of the company and need to figure out a distribution relationship. So these are journeys for us in the next year or two.
How do you manage geographical challenges for Edge applications?
The Edge applications today are serviced by well established companies, they have found their balance with their right kind of support and commercial infrastructure. But there is servicing in this from my perspective with old technology so we are bringing a infrastructure to re deploy and replace that but the support infrastructure is very pivotal and one key thing makes a difference which is software experience. If your software experience is easy to use and people can get more of a self-managed mode then the support infrastructure becomes dramatically less. So we have really chosen a more software centric customer experience.
What are the key focus areas and initiatives planned by SiMa for the next 12-18 months?
I have joked that I have only four problems. One is needing to build an amazing product, we need to have amazing customers, need to have amazing talents and have amazing investors. We feel very good and worked very hard to get into this position in the last two years and now it’s about being paranoid putting our head down every day and continue to do these four things very well. We are also well funded so the need for funding is for later.
Did the pandemic slow down the production?
I thought that the pandemic is really going to hurt us and if we look where we are now it has not. It comes down to the people we have and the resilience and their ability to adapt. I am amazed that we have hired more people in the company during the pandemic and we have never meet and we are working on a very complicated network from different geographies. I was very worried at the beginning of the pandemic as it will affect our ability to build a good product. Today the big shift is the reliance on new technology to balance the new normal, so people are building around the challenges.
What were some of the measures you initiated to manage the production?
What traditionally happen is people get in a room to get done but now its all remote. Product development is now completely remote and we have people globally working together. We have learned how to work with the technology partners and getting them give us the infrastructure needed for production. Hopefully in the next 6-9 months life starts to become normal the question is will we go back to the way we used to work or are we going to adapt to the new normal. So people are figuring it out how to work in this times, be in a room together socially distant, we have adapted we have not slowed down or changed anything and in some pockets the productivity have increased.
What are the measurable metrics for cost benefit ratio post deployment?
We have internally the concept of a ten x and that’s the aggressive goal we have taken is power savings or production cost or cost of ownership or machine learning ability, but each of this application is very tangible economic and technical merits and pass on to our customers.