It is not just India—most engineers working on AI, ML and data science are not specifically trained for the art. In 2018, The World Economic Forum suggested 54% of IT practitioners will need reskilling by 2022 to meet the requirements of AI. We are already in 2022, and today, experts deem, we will need upskilling as well as reskilling to excel in AI practice.

Analytics India Magazine got in touch with experts in the field to understand the challenges we currently face and how we can overcome them.

Introduce AI at a younger age

The first step to encouraging AI specialists is to interest them in the subject right from the students’ initial learning days. “Coding is being taught in schools at a younger age, and they are doing a good job”, noted Swati Jain, Vice President of Analytics at ExlService. She recalled AI competitions where the students won most prizes despite there being participants from the industry present.

“The mindset for AI has to be developed from school, (because) the ability to analyse starts from the core”, Swati said, stating schools don’t need to have compulsory programmes, but introducing it to students to explore in the future is the way to go. The Indian government’s National Education Policy (NEP) 2020 is a step in this direction of technology.

Specialised AI training

According to Kaushik Sanyal, Global Managing Director at Accenture, the biggest challenge is the fallacy that engineers are suitable to be AI technologists. So, to him, it didn’t come as a surprise that only 2.5% of engineers possess the real skills for AI, given they wouldn’t own AI skills in the first place. “Civil engineers should be civil engineers”, he said, arguing AI engineering needs to be taught as a specific program at universities. “(The problem in India is that) we don’t have many institutions offering such courses.”

Specialised learning is important because data engineering, data science and data visualisation are different fields that should be treated differently. “Data science is a combination of mathematics, engineering and analytics. While the former two are taught in schools, analytics is completely skipped”, explained Kaushik. This leads to practitioners having conceptual skills without the understanding of applying them. Another issue arising with non-contextual learning is the problem of the language. “The technical lingo keeps changing, and data scientists need to evolve with it, but unfortunately, schools do not cover the correct language needed.”

Even when reskilling existing engineers, they are most likely picking up lessons on the job. There are no bridge academic courses that teach technical AI skills as a top-up to engineering. “Not too many colleges offer special bridge courses that companies can use. Any engineer may adapt to working in AI, but he won’t know the principles of it,” Kaushik said. “India needs more institutions for AI, that have curriculums specifically designed for AI engineers, be it an undergraduate course, (postgraduate) or a bridge course.”

Internships build real knowledge

While universities are important, they may not teach AI engineers all they need to know. Computer science is a field where practical exposure is considered to be more valuable to build one’s knowledge of the field. Ashwin Swarup, Vice President of Data Science at Digite, holds a similar belief, asserting that doing internships while studying is the secret sauce to earning AI skills.

Many engineers are now working in the AI industry because of the ease of using online libraries or Kaggle to create projects. But as Ashwin asserted, “the challenge arises when companies have to appoint a person that can find the problem in the company, create a problem statement and suggest the right business solutions.”

The problem lies in the lack of practical teachings in engineering universities. “Finding the right AI/DS solution involves asking the right questions, creating a problem statement, having a business understanding and multiple domain knowledge to provide the correct business solution. Engineering does not teach this,” he explained, providing a solution- ‘Get internships. Engage in doing internships throughout the course of the study.’

Encourage research

Academia and research are integral to learning the subject in-depth, innovating, and keeping up with the developments, but “the value for research is very less in India. Barring a few institutions like IITs, most don’t focus on encouraging students to pursue research”, Ashwin noted. As a result, India is deficient in research, especially in comparison to international committees, where Indian academia falls short of competing for papers. Interestingly, Ashwin said, “The volume of research papers coming from China, the country we want to compete with, is overwhelming.”

This challenge can be overcome by two institutions, academia and companies. Indian professors fall short in their ratings in comparison to international instructors. India creates phenomenal students, but the Indian academia needs to amp up the experience of their professors and create an academic body that supports quality research. “Quality in papers is important, and it can be measured through the number of citations. Publishing research just for the sake of it will not do anymore. The quality is more important in (Indian) academia.”

Another opportunity is for companies to conduct research. At the international level, Facebook, Google, Uber, and more are known worldwide for their research and development. “International corporates are moving into academics. Indian companies are not keeping up with this”, Ashwin said. “The papers are open-sourced, so they get more ideas from the public and give back to them. You hardly see this in India.”

Upskill the workforce in adopting AI

According to Swati, Indians may not produce the relevant engineers, but they can surely upskill the present ones to meet the technical demands of AI. “Most importantly, the entire workforce needs to understand the value chain of analytics. Education of the workforce as a whole is needed, as not all companies can afford specialisation”, she explained. This is in line with the upcoming trend of preferring ‘generalists’ in the workforce. “Companies should focus on upskilling in a continuous fashion with a clear dedicated budget for capability development.” Lastly, companies should take an aggressive, systematic approach to even train the non-AI workforce on the technology. “We may only have 2.5% ‘relevant’ engineers, but with proper training, India can meet the demands of AI,” she concluded.

(This is a slightly modified version of an article originally published in Analytics India Magazine. The original article can be found at