India's IT firms face a new dilemma: How to train tomorrow's engineers in the age of AI

Indian IT companies are moving away from hiring large numbers of graduates and teaching them everything after joining. Employers increasingly want newcomers to understand technology, customer needs and real business problems from the start. This matters because AI is reducing some routine work that once helped beginners gain experience.

TCS and Infosys show the change clearly. TCS hired more than 110,000 engineering graduates four years ago, but its financial-year 2026 intake was less than half that level. Infosys recruited about 80 per cent fewer fresh graduates in financial year 2025 than three years earlier. Both companies now include generative AI in graduate training.

The change does not mean companies have stopped hiring beginners. Multinational firms still recruit them, but in smaller numbers and move them faster into roles combining technical and business knowledge. Universities are also being urged to teach problem-solving, decision-making and wider technical skills earlier. Without this investment, experienced-worker shortages could emerge later.

Generative AI is technology that produces new material, such as text, images or computer code, after receiving instructions. In software work, it can help generate code, suggest changes and support testing. This matters because many entry-level tasks were repetitive and helped junior engineers build experience before moving into more complex work.

The article reports that Infosys co-founder Nandan Nilekani said AI was increasingly taking over coding tasks traditionally handled by junior engineers. Oliver Wyman’s Sumit Sarawgi also said routine coding and testing were declining. AI can complete or accelerate these tasks, leaving engineers to focus more on customer problems, business transformation, design and managing technology.

This shift raises both opportunities and risks. Graduates may become productive faster if they learn to use AI well. However, they may get fewer chances to practise basic work on the job. The article says true specialisation, such as becoming a machine-learning engineer or AI architect, still requires years of hands-on experience.

The reduction is substantial, but the companies report it in different ways. TCS hired more than 110,000 engineering graduates four years before its financial year 2026 intake. In financial year 2026, TCS hired fewer than half that number. The article does not provide an exact latest TCS total.

Infosys reported a sharper percentage comparison. It recruited about 80 per cent fewer fresh graduates in financial year 2025 than it had recruited three years earlier. That means its intake was roughly one-fifth of the earlier level, based on the percentage given in the article. The comparison concerns fresh graduates, not all employees.

These figures reflect changing workforce needs rather than a complete end to graduate hiring. Companies are placing greater value on experienced workers and applicants with practical technical and business skills. TCS and Infosys are responding with graduate programmes that include generative AI. The longer-term concern is whether fewer beginners will create enough experienced specialists later.

Coding ability alone is no longer the main measure of graduate readiness. Employers increasingly want people who can understand a customer’s problem, identify the business context and work backwards to a useful technology solution. This matters because AI can handle some coding, while human value shifts toward judgment, communication and applying technology effectively.

The article highlights growing demand for AI, cybersecurity and process improvement. It also points to client needs, business transformation and technology management. Indian universities are being urged to develop problem-solving, decision-making and broader technical skills earlier. Candidates such as Arpita Dayama are building data-science projects and working with agentic artificial intelligence to show practical ability.

The strongest graduates may therefore combine technical depth with business understanding. They must explain their work simply and discuss technical details when questioned. Companies are also moving graduates more quickly into roles that require both technical and business knowledge. This raises the bar for entry, but it reflects work that is becoming more complex and less routine.

The central risk is a missing middle of experienced technology workers. Entry-level employees normally gain knowledge gradually, then become specialists, team leaders and managers. If companies hire fewer beginners, there may be fewer people developing through that pipeline. This matters because AI, cloud computing and cybersecurity roles already require expertise and quick deployment.

A 2026 Zinnov survey found that about 60 per cent of multinational companies’ centres in India preferred experienced hires for these areas. Kamal Karanth said training can make junior employees comfortable using AI, but true specialisation as a machine-learning engineer or AI architect takes years of hands-on experience. Training alone cannot instantly create that depth.

Companies could face higher competition for experienced workers and fewer future middle managers. Oliver Wyman’s Sumit Sarawgi asked where future middle managers would come from and urged investment in graduates. The article points to a longer-term solution: continue developing young talent, even while hiring fewer people and automating routine work.

Companies and universities are trying to replace some of the practical learning that entry-level jobs once provided. Employers are adding structured training, while universities are being asked to teach broader technical and problem-solving skills earlier. The goal is to make graduates useful sooner without abandoning the development of young talent.

TCS says its graduate programme emphasises hands-on learning and includes generative AI. Infosys includes generative AI while focusing on problem-solving and design thinking. Students such as Arpita Dayama are building data-science projects and working with agentic artificial intelligence. Employers also want candidates to explain their projects clearly and connect them with real customer or industry needs.

These approaches can improve readiness, but they have limits. Training can help junior employees become comfortable using AI, yet specialists such as machine-learning engineers and AI architects still need years of experience. Companies therefore need closer university links and sustained investment in graduates. Otherwise, faster preparation may not replace the depth once gained through work.

Beginner jobs matter because they provide repeated practice and a gradual route into complex work. Junior engineers traditionally handled coding and testing, learned how technology served customers, and developed judgment over time. That experience helped companies build future specialists, team leaders and middle managers rather than relying only on people hired from outside.

AI is changing this pathway. Infosys co-founder Nandan Nilekani said AI was taking over coding tasks traditionally handled by junior engineers. Oliver Wyman found that about 45 per cent of managers in India said AI was taking over tasks that traditionally helped junior employees gain experience. As routine work declines, beginners may have fewer opportunities to learn by doing.

The industry now faces a development challenge. Companies still need people with deep skills, but true specialisation takes years of hands-on experience. Oliver Wyman’s Sumit Sarawgi asked where future middle managers would come from. Graduate training, university preparation and carefully designed early-career roles may be needed to preserve that progression.