This article is a part of a series of articles. The first part of the article can be read here. The gist of the first part was that there is a need to understand the skill levels of today’s youth in India and then reconcile it with the skill requirements. This needs to happen before a reskilling strategy can be formulated. However, doing it practically in a country with a population the size of India is extremely challenging. The good news is that Generative AI can help by being a solution component, that can help understand the skill gap.

The starting point of any analytic solution is to develop a high-level blueprint of what you are trying to solve. That high-level blueprint starts with a very simple logical relationship. For our specific use case, the underlying problem is to match supply with demand. The need for reskilling arises because we believe that there is a pool of people who can be skilled with certain skill set, and then matched with the jobs pertaining to those skill sets, that currently can’t be filled. So what we want to do is basically build a supply of skilled resources that can match the demand, which is in the form of job openings across industries, throughout the country.

Let us start with the relatively easier aspect first. That aspect is capturing the demand side. When I say it is relatively easier, I mean it is relatively easier as compared to the supply side, as we will see in the later parts of this article.

We have to understand that while many industries and companies within these industries in India are organized, there is a huge unorganized sector as well, which also evolves with time and hence the skill set required for jobs in these companies also evolve.

Why is this important? You have to remember that we are looking to explore leveraging generative AI. For devising a reskilling strategy we need some form of data source to train a model.

When it comes to organized sectors, the job openings are generally documented. By training a model, we can extract the key skill sets from these job openings. Hence, the very first step obviously is to create a massive data set in collaboration with organized companies across industries. Many large organizations or multinational companies operating in India already have decent job descriptions across job boards. That becomes a straightforward source of data. However, just to make sure that all the data is being captured, the government can float an initiative and ask these companies for data. The companies will be more than willing to participate to contribute data in the form of skill sets that they need.

In order to make sure that we capture the data in an organized way, a good first step will be to create a template to capture skill data by company and by industry. A template that can be used to collect this data directly from the companies, as well as to extract this data from other sources like job postings on the Internet into that template. This template hence becomes a very crucial element of the entire process. It is also a good example of why I keep on emphasizing that processes and people are as important as technology. Let us assume that we are able to capture this data in this format.

The challenging part that then comes into play is the unorganized sector, as identified above. These are jobs that go beyond plain manual labor and involve some form of skill sets. For example, a small industry making cushions for chairs may need people that have experience operating certain technical machines. These machines evolve over time and hence the technical dexterity needed to operate these machines evolve as well. Since there are millions of such small unorganized companies across the country there are millions of opportunities.

For example, I was reading an article that stated that many of these small industry owners in the trans-Yamuna area in Delhi have been facing acute scarcity of people who have the aptitude to learn how to operate such machines. So while every day trains full of manual laborers from states like Bihar land in Delhi, the skill that is needed is still in scarcity. Many of those who arrive, some with high school education, can easily learn these skills but have no idea that there is a demand out there. That is why capturing these skill sets in unorganized sectors is also crucial. Unfortunately this will be a very manual process, and perhaps can be pursued through associations of these small business owners to collect data in bulk.

Now let us assume that we have been able to create the data source. If we were able to design that template to capture data in a strategic way, we then start building and training our model. Essentially, what the model first needs to learn from the demand side perspective is a comprehensive list of a combination of skill sets, associated expertise, educational background, etc. Next, it captures what are some of the most crucial skills in demand, based on the list, impact on economy and then assign criticality scores. Remember, chances are high that the Pareto principle will come into play, where 20% of skill sets will account for 80% of open jobs. This analysis will obviously be made much more granular by the model since that granularity needs to be matched with the supply side.

The important aspect of this model will be that it will be trained solely for this purpose. That will allow it to keep learning over time and fine-tuning the process of understanding what are the key skill sets that are needed to keep the economy driving. But these job descriptions and template skill requirement will not be the only data sources that this model will use. In the next part of this article, we will explore more details pertaining to this model.

The next part will be published on 13th of July.