In the tug of war between India’s IT employees and employers, workers are at a loss.

The landscape of the sector is changing rapidly. Visa pressures in the US and in other countries have compelled Indian companies to ramp up local hiring abroad, putting Indian professionals out of the running for roles they would have easily bagged before.

Additionally, the move away from cookie-cutter software jobs, and towards specific hi-tech services in areas like artificial intelligence and machine learning means that IT companies are no longer looking to increase their headcounts. Instead, they want to foster a smaller, more niche talent pool.

In response, Indian IT professionals are seeking out other places of opportunity or scrambling to unionise, but these are short-term fixes. There’s only one sure-shot way to ride out the continuous waves: re-skill.

At Infosys, the country’s second-largest homegrown IT behemoth, the focus is on teaching employees to adapt quickly to changing times. Quartz spoke with Navin Budhiraja, the head of architecture and technology at the company, about what’s going on in the new age of learning in IT.

(The conversation has been condensed and lightly edited for clarity.)

What changes must the tech workforce undergo to keep up with the industry?
The pace of change is very different from how things have been in the past. It is becoming incumbent for everybody to create this workforce, as Vishal [Infosys CEO Vishal Sikka] calls it, “full of innovators”. This cannot be just a chosen few in the company. It has to be kind of grassroots because everybody is getting impacted... It needs to happen in all parts of the company. People who work with customers, people who work in operations, and so on and so forth.

How important is it to train your employees in specific upcoming fields like artificial intelligence and machine learning?
If you focus too much on specific technologies, you kind of lose sight of what you’re really trying to achieve here. The focus for us at Infosys is evolving from learning specific technology skills to learning to learn. Whatever you learn today is likely to change and evolve very quickly over time.

We used to teach (the students) Java or .NET or whatever it might be. (But) one of the very early things we started doing a couple of years ago is to teach them three languages at a time… It’s not about just teaching the particular skill. It’s really about how these things work, how you apply them to a particular case.

How do you simulate these real-world applications?
It’s important to see how you apply the concept you have learnt to a real use case. How do you deal with failures? We have [developed] 20 very compelling use cases and real programmes, and we take that as a starting point for somebody to learn a [computing] language. We bring experts who can actually answer questions... Inside the classrooms, we actually get them to collaborate and do team projects early on because that’s how it will be.

Has hiring for the new centre in Indiana posed any challenges?
We have not had a problem hiring the people. Our focus is to make sure that these people have the right ability to learn, right amount of passion, and they come with a certain set of skills (so) that they do a good job.

Back in India, is there truth to reports about the quality of engineers being poor?
There has been recent talk that they may not be ready for the kinds of jobs that they’re expected to do. But in some sense... these reports miss the point. Some of the measurements they were doing were specific to skill sets or technology, like the programming language. My feeling is even if they have succeeded in doing very well on the specific assessment, I don’t think it will tell you whether the person or set of people are able to do the job you expect them to do.

Measurements going forward should test how good a person is at finding a problem and learning the skill set they need to solve that problem.

How important is it to hire from top tier institutes like the IITs and MITs of the world?
We have always hired from the breadth. It’s possible that we get a spectrum of people on some basic skills. Most of them, I still feel, need to come in and really need to understand how they use the skills and solve problems. That’s why there’s this big focus on training and education.

What’s extremely important is that you really understand your end user. In the past, there was more focus on the tech aspects, [like] what features should be there, and maybe not as much on how you deliver that or how it is experienced by the end user. The second aspect is that if you look at areas like AI and machine learning, it’s really important that you understand the domain, what parts of the data are needed to solve the problem. Let’s say you want to solve a problem in medicine or around insurance. What is really needed is to start with what the problem is, what the end user’s challenge is. We are even bringing in people from these different areas, people from non-STEM fields as well, to bring perspectives [and] end-user empathy.

How real is the threat of automation making some tech jobs obsolete?
Automation replacing or taking away many of the jobs that exist today, that is a reality. One has to assume that [the repetitive] jobs over time will be automated. We tend to focus on [what] we only know about things that exist today. What is clear is that we have to focus on jobs that don’t exist today.

Let me give you an example. We all thought [choosing to work in] farming was something that went away 50 to 100 years ago. But we have seen a big resurgence. There’s a lot of work going on using tech; how do you do things like vertical farming? One would not have imagined even five years ago that these are the jobs our kids want to do.

There may be some short term-disruption [because of automation] but the focus needs to be on how you use tech, on how you improve on what exists, and, more importantly, to find things that don’t exist yet.

This article first appeared on Quartz.