Artificial intelligence has generated both excitement and anxiety among students, fresh graduates and mid-career professionals, particularly in software development. Coding, built on structured logic and repeatable patterns, has proved especially amenable to AI training. As models grow more capable of generating, testing and even debugging code, many are asking a blunt question: what exactly should a software professional now learn to remain relevant? Vishal Chahal, VP at IBM India Software Labs, argues that the answer is not to compete with machines on speed, nor to abandon programming fundamentals. It is to elevate one’s thinking. “AI is redefining software. Coding is only one part of the software life cycle.” Design, architecture, deployment, support and continuous improvement remain firmly in human hands. “AI is redefining what you will do in your job (as a software developer)”.The productivity gains are real and this has implications. In Chahal’s own experience, developers can see “at least a 30 percent uplift in daily coding tasks”. Essentially, software can be built “much faster”, test cases can be generated more quickly, and iteration cycles have shrunk. This means teams can now experiment more freely because they can fail fast and try again without the same cost in time.
.
The deeper shift lies in how engineers think. “Instead of spending your time writing code line by line, you should be thinking about systems,” he says. “How am I designing this system? What solution am I trying to achieve?” Once that clarity exists, structured prompts can guide AI tools to generate much of the code. For students, the message is not to discard programming languages either, even if AI can write much of the code. “We must continue to learn the fundamentals,” Chahal says. Understanding how software interacts with hardware, how prgramming languages translate into machine instructions and how systems behave under load remains essential. AI-generated code still needs to be understood, validated and improved.However, “coding itself is no longer the super skill,” he adds. “The super skill is the ability to take a requirement, turn it into a solution, and then express that solution clearly through structured prompts.”Vague instructions will yield vague results. “If you say ‘write a good JavaScript program’, that means nothing. You must define what ‘good’ means — secure, efficient, scalable, compliant. You must specify the constraints.”Chahal cautions strongly against intellectual complacency as well. “If you offload all your thinking to these tools at the start of your career, you will not develop the ability to design complex enterprise systems.” Building prototypes with AI is one thing, designing mission-critical digital infrastructure is another. Architectural judgement in this space is built through understanding gained over many years at the workplace.Security and governance, he argues, are also now becoming foundational skills. With AI generating code and developers pulling from open-source repositories, risks multiply. “You must know how to build secure, governed solutions,” he says. Engineers should be able to scan for vulnerabilities, detect data leaks and apply responsible AI principles.Chahal rejects the idea that entry-level roles are disappearing. “Jobs are not going away. They are transforming,” he says. Which is why Chahal and his team at IBM now look out for candidates who are adaptable. “The hunger to learn and the ability to unlearn. That’s what we look for.” he says. Linear, narowly defined career paths matter a lot less than evidence of flexibility — shifting domains at work, learning new tools and embracing change. Chahal’s advice is to try and highlight these competencies in your resume if you have them.For both young and mid-career professionals, his other bit of advice is to practice daily. “Spend half an hour or an hour every day using these AI tools.” The objective is to be intimately familiar with them — to understand the nuances of these AI models, their limitations, and their rapid evolution. Those who stay close to the change will recognise the shifts between one wave of models and the next and will be able to adapt in a cutthroat jobs marketplace more effectively because of it.
