Why Indian IT is Quietly Abandoning the Pyramid Workforce

For decades, Indian IT services scaled on a broad base of junior engineers supporting a narrower layer of experienced architects and managers. That pyramid hierarchy was designed for effort-based billing and predictable delivery, fuelling steady campus hiring.

That structure is now being quietly reworked, not through formal announcements, but in how work is staffed, priced, and delivered.

During IT services company Mphasis’ third-quarter post-earnings press interaction, chief executive Nitin Rakesh said the industry’s long-standing focus on the workforce pyramid is becoming less relevant as AI changes the underlying economics of delivery.

“The pyramid is actually an old-school question,” Rakesh said. “It doesn’t matter. It’s about having the right skills, complemented by the right software solution, to drive an outcome.”

Automation at the Base

The bottom of the pyramid is the most affected. Tasks traditionally handled by junior staff—basic support, repetitive testing, and routine execution—are increasingly automated or heavily AI-assisted.

Rakesh described that in Mphasis’ delivery stack, L0 and L1 activities operate without human intervention, while higher layers are becoming progressively automated.

“L0 and L1 are completely human-less,” he said, adding that L1.5 is “highly automated” and L2 is supported by AI tools, including developer and operations assistants such as Devin.

This shift weakens the widely-followed strategy of having large, junior-heavy teams. “We are not doing capacity-based pricing,” Rakesh said. “Whether I need the same number of people, half the people or one-third of the people, the client has no consideration. What the client cares about is the outcome.”

He also pushed back on the idea that AI will allow IT firms to charge a premium simply for domain-specialist talent. He said domain knowledge is increasingly embedded into the delivery stack itself through platforms such as Ontosphere, Mphasis’ domain-contextual knowledge layer, rather than sold as a separate skill.

“We’re not charging for skill anymore. We’re charging for a solution,” he said.

Rakesh explained that Ontosphere is designed to encode domain context, whether in banking, payments or insurance, directly into the knowledge graph and orchestration layers that guide delivery. As a result, solutions already combine domain architects, enterprise architects, AI agents, and human teams, along with ecosystem partners where required.

“This is an ecosystem plus a human play,” he said, adding that hyperscalers and platform partners form part of the stack depending on the client environment.

As AI becomes embedded across delivery, Rakesh said pricing models are shifting away from individual skills toward bundled solutions and outcomes. “We’re not pricing for skills as much as we used to. We’re in transition with that.”

What “AI-Infused” Delivery Really Means

In a conversation with AIM, Gaurav Vasu, founder of cognitive intelligence platform UnearthInsight, notes that the term “AI-infused delivery” covers a wide range of activity, and only part of it truly disrupts organisational structure.

At one end, AI speeds up traditional work: developers use copilots to write code, testers auto-generate test cases, and support teams summarise incidents. The scope remains the same, but fewer people are needed to deliver it.

In the middle, AI appears as an embedded feature—chatbots or automated summaries layered into existing programmes—without changing pricing or governance.

The real inflection point, Vasu notes, is when AI becomes core to engagement. That includes operating AI agents, orchestration layers and decision systems that require governance, auditability, and human oversight. “That’s when delivery stops being about effort hours and starts being about system behaviour and accountability,” he remarks.

Those programmes need fewer junior executors and more people who can supervise, validat,e and manage AI-driven workflows.

Fewer Juniors, Faster Ramps

The impact on talent is already visible. Rakesh said Mphasis has moved away from traditional campus hiring over the past two years, replacing interviews with internal programmes and hackathons focused on live project work.

“Number one skill we are looking for is learnability and technical skills,” he said.

This does not mean early-career hiring is disappearing altogether. Rather, it is becoming more selective and skill-specific.

Neeti Sharma of TeamLease Digital observes that companies are paying a clear premium for freshers and early-career hires who can work inside AI-enabled delivery environments. Roles such as “specialist programmers”, which are in high demand with some firms paying up to ₹22 lakh for entry-level positions, are expected to design AI workflows, integrate models into enterprise systems, and manage data retrieval and governance. The premium, Sharma has notes, is not for theoretical AI knowledge, but for applied skills.

Most freshers, by contrast, still enter for support or basic development roles—precisely the areas where AI is compressing demand.

Pressure Builds in the Middle

As the base narrows, pressure is shifting to the middle of the organisation. Engineers with 5–10 years of experience are increasingly expected to supervise AI-assisted delivery rather than execute tasks manually.

UnearthInsight estimates that while only a small portion of revenue generated from large deals is directly attributable to AI work, a far larger share of delivery effort already uses AI tools internally. That gap explains why firms are reluctant to release any breakdown of AI revenue—the impact is showing up more in productivity and margins than on invoices.

Clients, meanwhile, are funding AI not as experiments but as targeted upgrades: document automation, legacy modernisation, customer operations, fraud detection, and IT infrastructure optimisation. These projects reduce billable hours and turnaround time, directly reducing the need for large junior teams.

Pyramid, Diamond, or Something Else?

Industry leaders differ on what replaces the pyramid. Some expect a diamond-shaped structure, with fewer juniors and a thicker middle. Others caution against eliminating early-career hiring altogether.

In a recent LinkedIn post, Ashwin Mittal, executive chairman of B2B analytics solutions provider C5i, argued that organisations which stop hiring fresh graduates risk running out of future leadership. He suggested a “rocket” model instead, using AI to accelerate early-career learning so younger employees reach productive roles faster rather than being phased out.

However, Rakesh suggested the end state remains fluid. “The pyramid will have some shape to it,” he said. “It’s probably more likely to have a little bit of a diamond shape.” 

However, he added that such transitions take time.