Is Talent Crunch a Spoiler for India's AI Industry? The war for AI talent, henceforth, would be ruthless enough to easily dwarf the challenge for spotting good software engineers

Aiming for market disruption with your artificial intelligence (AI) solution? Spoiler alert! There is less than just a handful 10,000 number of specialized talent in AI in the entire world. Baffled? The war for AI talent, henceforth, would be ruthless enough to easily dwarf the challenge for spotting good software engineers. As it happens, you know that machine thinking has begun to usurp human thinking. Despite India's global dominance in offering cheap engineering talent at scale, young AI companies in India are looking for alternate ways to suck in what's available.

A mid-year 2017 survey by recruitment platform Belong. co tries to read between the lines and throws up some startling numbers. Particularly for data scientists wherein the essence of AI skills lies - ranked 0.8 on Belong's talent supply index. This, as per Belong, indicates of only 800 data scientists available for 1,000 data scientist opportunities, which means significant mismatch in demand and supply of AI talent in the country. The other two figures that drives home the point of talent crunch includes just four per cent tech talent with expertise in deep learning and neural networks system and "less than two per cent having a PhD degree in AI-related technologies," says Belong. And to develop great AI products, that's how premium talent and expertise should be - both quite scarce in India.

Another Way Out

Globally, all good AI talent goes to the big four -- Amazon, Google, Microsoft, Facebook that include inbound hiring, hiring from colleges, poaching it from each other and rest of the big tech companies, and acqui-hiring young companies. For e.g., Google acquired Bengaluru-based AI start-up Halli Labs in July this year and Hyderabad-based TupleJump was acquired by Apple last year. Other big and small tech companies are left to pick from people who have basic understanding in deep learning with requisite math skills and then groom them over few months through training and development.

Ratan Tata-backed AI-based personal assistant app Niki.ai looked everywhere to get hold of some good AI talent but nothing worked.

"Indians have very good software skills but not many of us are exposed to industry applications around machine learning (ML). So I realized that we will have to look for in-house talent development by hiring out-of college kids and grooming them," says Sachin Jaiswal, Cofounder and Chief Executive Officer, Niki.ai. Jaiswal along with three other co-founders (his batch mates from IIT Kharagpur) had also hired three interns from the US, UK and China. Out of its 42 headcount, 34 are engineers working on its AI platform.

Apart from hiring them nascent and training them, start-ups which are uniquely positioned in the market have been able to attract talent quickly. From doing menial coding job in the back-end of an e-commerce company, they have been put in the driving seat to create great AI products. "AI talent is scarce in India because it is located at the wrong place. They have to be leading the front-end of innovation which happens in the Silicon Valley," says Ravi Shankar, Co-founder and Chief Executive Officer, Active.ai. The start-up launched last year by Shankar - former Group Executive Vice President at Yes Bank, offers conversational banking services to banks and raised $3 million from Kalaari Capital and IDG Ventures.

Active.ai, claims Shankar, is one of the very few and unique financial services company applying AI in the front-end or customer facing, which has helped him attract lot of talent. Word of mouth has also worked.

"Earlier it was a challenge but now we have hired at every level. There are young people from Universities like Stanford, Massachusetts Institute of Technology and Indian Institute of Technology (IITs) doing summer training in ML and AI. There are also experts who have authored books on AI etc.," he adds.

Investors too are high on backing companies which have an AI element in their business. Rajesh Raju, Managing Director at Kalaari Capital (among the most prolific AI investors in India) affirms the challenge of manpower supply in AI is because of having massive opportunity led by the technology.

"We see talent crunch every day because most of our companies are run by data scientists. We want to make sure that every business which pitches to us is using AI, ML etc., and have in-house talent to not only gather data but also connect the dots between the data and make algorithms better," says Raju. Among the AI companies backed by Kalaari Capital include Vernacular.ai, Embibe, Active.ai etc.

This is akin to every innovation and new technology for which supply is succeeded by the demand. We can look at it in two ways - first, building the core processing system of neural network and second, its application for commercial usage.

The former is built by the four tech heavyweights in the Valley and this is where the real crunch is. "The talent crunch is in terms of speed at which these companies want to build or transform all the systems into AI systems or neural networks. You won't find many trained programmers to do," says Milan Sheth, Partner & National Leader - Technology Sector, Advisory Services, Ernst & Young India. For its application, Sheth doesn't see a problem in hiring globally.

Coming back to the ways, referrals compared to any other business helps in locating hidden talent which might not be accessible otherwise through traditional hiring route. That's much like Avneesh Agrawal who runs Netradyne that offers driver safety platform for commercial vehicles called Driveri. "Talent attracts talent and in our case 80-90 per cent of it is referral based." The company has centres in Bengaluru and the US where it has several PhDs in AI apart from junior talent. "We look at people while hiring who are technically strong, great aptitude, ability to go after unknown issues and do research," adds Agrawal.

Throwing Money Around

As a global problem, talent crunch has pushed tech companies to award AI talent with astronomical pay cheques so much so that in the US alone, companies are spending a whooping $650 million or above in annual remunerations this year for 10,000 jobs related to AI, as per hiring data and salary tracking firm Paysa. For Jaiswal, 40 per cent of his cost goes into grooming talent even as he doles out typically twice more than what a software developer gets usually.
"A top tier talent from any of the engineering colleges in India gets paid pretty much similar to that in the US which can be around $100- 120k. But these kids go to the US because they want to work amidst best engineers in design. Even for an above average talent in the US, $50- 60k has become pretty much standard," explains Shankar.

While Indian startups didn't disclose the exact salaries they offer to data science talent but according to outsourcing and consulting services company KellyOCG, what Indian companies offers (box below) is in high contrast to what the US companies pay - Rs 1.9 cr - 3.2 cr per year.

Losing on Academia

Then there is also a challenge at the academia level. While institutes like IIT Hyderabad, IIT Madras, IIT Delhi, IIT Kanpur, IIT Hyderabad, S. P. Jain Institute of Management and Research, IIIT Bangalore offer specialized programs related to ML, data science, data analytics, neural networks and AI but they are doubted for being out of sync with industry requirements.

"In India, the talent crunch is because the academia hasn't adopted the latest developments in data science in their curriculum. There are people with legacy knowledge but aren't updated with modern techniques," says Navneet Sharma, Co-founder and Chief Executive Officer, Artifacia - Bengaluru-based visual discovery platform for fashion and lifestyle retailers.

But professors at IITs vehemently disagrees. Associate Professor Mausam at IIT Delhi's Department of Computer Science and Engineering teaches specialized course on AI to more than 150 students annually. "We teach the most advanced technology since we understand the area as researchers much more intimately than being taught from a textbook," he says adding, "Of course certain learning is needed as an engineer going in an industry."

In comparison to the Valley, apart from its institutions, the large tech companies also have set-up AI labs inside or near institutions to stay in proximity to the local talent. For example, Google's parent company Alphabet earlier this year announced setting up of its research facility Google Brain in Toronto near universities for access to researchers and young graduates. Similarly, Amazon increased capacity of its research and development center in Cambridge, England with additional 60,000 sq. ft space for more than 400 scientists and engineers.

Today, the talent hiring in AI is centred around solving for one use case at a time, which is, focusing on specialized AI solutions to understand how customer adapts to a particular AI instead of general AI solutions that can multi-task. For instance, WhatsApp knows what the next word you would type and it shows that since it uses AI.

But if you type for e.g. 'I would pay Bill' wherein Bill could be a person or an invoice then WhatApp won't know what you mean by Bill. So AI has to learn to differentiate between the two. "It has to go hand in hand with the customer expectations," says Shankar. Hence, more talent doesn't lead to a smarter AI.

Moreover, 40 per cent of India's data scientists are involved in IT (services) followed by computer software and financial services with 28 per cent and 17 per cent respectively. Apart from actively scouting in these sectors, AI talent would be tough to come across even on LinkedIn. Nonetheless, scrolling through research authors, patent holders in AI related areas and building own your technology background to attract possible talent in your network might find you good AI prospects if not the best talent.

(This article was first published in the December issue of Entrepreneur Magazine. To subscribe, click here)