LINEAR SQUARED SIGNALS AGGRESSION ON GLOBAL EXPANSION

The tech startup specializing in artificial intelligence and machine learning has hired former Google Cloud Country Manager for India Mohit Pande to drive expansion in Southeast Asia and the Middle East

Linear Squared, a technology startup specializing in developing sophisticated algorithms for artificial intelligence (AI) and machine learning platforms, sees an opportunity to grow the business globally, but finding the right talent is challenging.

In 2017, the company developed an AI solution to boost productivity at apparel manufacturing plants and save up to $30,000 in costs. It was hailed as a world-first by Google India Vice President Rajan Anandan, who became Chairman at Linear Squared when the venture capital firm BOV Capital he co-founded invested in the startup.

The startup has since ventured into building complex platforms incorporating AI and machine learning for Sri Lankan businesses in banking, FMCG and telecommunications, but local demand is not growing fast enough.

The company recently hired former head of Google Cloud India Mohit Pande, whose new role as Vice President at Linear Squared is securing new markets in the region.

“Global spending on machine learning platforms alone is around $12 billion, and this is expected to increase to $60 billion by 2021. Linear Squared is well placed to capture this growth and we are looking at entering markets in Southeast Asia and the Middle East,” Pande says. However, the company is challenged by a talent crunch. The country’s university system doesn’t produce enough data scientists required to write code for AI and machine learning platforms, says Sankha Muthu-Poruthotage, chief executive and co-founder of Linear Squared. “As a country we need to have an open mind about making it easier to hire foreign professionals,” he says.

Excerpts of an interview:

Sankha Muthu-Poruthotage, chief executive and co-founder of Linear Squared, believes the Sri Lankan startup is poised for massive growth by venturing into new markets, but to do this, it must bridge the talent crunch

Can you give us a sense of what the opportunity is like expanding into new markets overseas?
Pande: Global spending on machine learning is around $12 billion, and demand will increase to $60 billion by 2021. This is an opportunity for Linear Squared to grow the business not just here in Sri Lanka but in the region as well. Big businesses here are already investing in AI, Cloud and machine learning but others too will begin investing in these technologies as they grow. We have a growing base of Sri Lankan enterprises and we also want to expand regionally to Southeast Asia and the Middle East. We are also looking at aggressively hiring data scientists and machine learning specialists. I think Linear Squared has all the makings of a global company which has started out of Sri Lanka. The problems that Linear Squared is trying to solve are common to businesses wherever they may be. We will continue to build solutions for businesses in the areas of apparel manufacturing, construction, food and consumer retail, financial services and telecommunications. It’s a great time to be in the tech startup world. Data sciences have evolved primarily due to the accessibility of fast and affordable computing platforms. We are able to solve problems in ways we never imagined. In that context, I see huge potential for AI and machine learning and Linear Squared plays right to it.

What is your expected growth over the next couple of years?
Muthu-Poruthotage: I expect revenues to grow at least five-fold over the next five years and this is just by playing to domestic demand. Growth will be much higher if we can capture business in Asia and the Middle East, and this is the next natural step for us. Some multinationals operating here already use our solutions, so there’s opportunity to leverage on those as well to reach global markets. With time, we expect revenue from overseas markets will outweigh the domestic market.

What are the challenges to achieving these goals?
Muthu-Poruthotage: We are probably the largest data science team in the country. Twenty out of 25 people at Linear Squared are data scientists and we want to grow that number. But finding enough talent for growth is a challenge. We don’t produce enough data scientists and software engineers who can deliver what is required, and many in this limited talent pool migrate overseas. One of the biggest challenges is to hire the right people and quickly. We don’t have a steady pool of specialists in high-end tech in order to expand. As a country we need to have an open mind about making it easier to hire foreign professionals so that domestic tech businesses can grow by capturing global markets.

Pande: Sri Lanka can learn from India’s example on how to grow the tech industry. India did a lot to improve the regulatory environment which attracted investments from global tech companies that opened offices there. They invested to develop infrastructure to support these businesses and also invested in developing local talent. Government policy on the Internet saw the online population boom from 100 million to 400 million, which created opportunities for startups.

Linear Squared has launched a platform incorporating AI and machine learning for apparel manufacture. What else are you working on?
Muthu-Poruthotage: We expanded into the finance sector, which I think is important. There are many applications of artificial intelligence and machine learning in the financial domain. However, Sri Lanka has been slow to adopt these, but demand is growing. Recent regulatory changes in the banking sector are forcing banks to adopt machine learning and predictive modelling, so we’ve seen demand picking up for our services.

For example, IFRS 9 is a new global accounting standard governing lending institutions like banks requiring provisioning for credit resources. Earlier, banks based their provisioning on how loans performed in the past. This is one reason for the 2008 global financial crisis; financial institutions were basing a lot of their provisioning on historical data rather than looking at future scenarios and outcomes to determine the probability of borrowers defaulting on loans. Since then, global financial market regulations have changed. Now,  IFRS 9 requires lending institutions to make provisions on loans based on predictions about future performance of borrowers. To do this, banks must adopt predictive modelling to come up with the probability of default not just of existing loans but future ones too. We’ve developed an application that will allow banks to do this and its gaining traction here.

We’re also doing some work in the area of credit underwriting. We are training machines to learn how to identify and predict the credit profiles of people so that lending institutions will know who to lend to or not. Machine-based underwriting is better because it eliminates arbitrary calls and subjectivity that humans can bring to the process. Over time, machines will be better at these predictive models than humans. There are many examples globally where financial institutions have been able to reduce their non-performing asset rates by adopting machine learning modelling. So these are some of the things we’re rolling out here.

We’re also working with food and beverages companies, a growing market in Sri Lanka as incomes improve and tourism booms. Businesses will continue investing in this space, including global food chains.

To give you an example of what Linear Squared is doing, we’ve developed a platform for a pizza store that helps them identify lapse. Pizza used to be a luxury but that is no longer the case. Now it’s a competitive business in Sri Lanka. We’re identifying individual customers who may no longer be patronizing this pizza chain. Some purchase weekly and there are irregular customers making a purchase every three months or just twice a year, so identifying customer lapse before it actually happens can be tricky. Monitoring market trends is not an effective strategy to identify lapse. Our model analyses trends of regular and irregular customers and identifies early warning signs of lapse so that the business can target their marketing.