As India’s GCCs take on global products, we have to hire and grow engineers who can make decisions.
Consider a payments team in India introducing an AI-based fraud capability for customers across several markets. If the model misses fraud, customers and the business lose money. If it blocks legitimate payments, customers abandon purchases and merchants lose sales. At the scale of global payments, a small shift in that balance touches a very large number of transactions. Someone has to decide where the balance sits, and increasingly that decision is made by teams in India.

The shift shows up in the research. Zinnov estimates that 70–80% of Indian GCC work in 2015 was execution-led. By 2026, expertise-driven and frontier work made up about 45% of the average India portfolio.
The mandates of newer centres offer further evidence. Nasscom and Zinnov report that 96% of GCCs established after FY2021 launched with a product or portfolio mandate. Centres are moving at different speeds, but the direction points towards ownership.
AI adds urgency. As tools take on more of the implementation, engineers contribute through problem definition, validation and decisions about what should reach customers. Teams responsible for global products need people who can connect a technical choice with what happens when someone uses the system.
Engineers have to understand the business before they can decide what to optimise.
Back to the fraud capability. Understanding the business helps an engineer decide what to measure. Alongside overall accuracy, the team might examine whether legitimate customers in a particular market are being blocked disproportionately. That can reveal a problem an aggregate number obscures.
Those business questions lead directly to technical choices. Does the training data represent the markets being served? How quickly must a decision arrive? What information can the model access? The team has to examine acceptable errors, security and integration with existing systems. AI-generated code in that pipeline deserves the same scrutiny as code written by hand. Reliability has to shape design, testing and deployment from the outset.
Good engineering decisions need agreement across functions and time zones.
Risk teams may favour tighter controls, while product and operations teams see the effects on customers and support volumes. Engineers have to explain the consequences of each approach, question assumptions and help agree on a workable balance. That takes technical depth, business context and the confidence to raise a concern early.
Global delivery raises the bar for clarity. The capability may serve markets with different payment patterns and rules, supported by teams in several time zones. Written decisions and clear dependencies keep everyone aligned. After launch, the same engineers monitor performance, investigate failures and revise their choices as evidence changes.
We have to build this capability across the whole workforce.
In a survey of 200 senior GCC executives, PwC India and FICCI found that 72% of centres needed to reskill or upskill more than 40% of their workforce over the next three years. Experienced engineers, architects and managers are central to that effort, because their judgement shapes architecture choices, reviews and delivery practices.
A backend engineer learning to bring AI into a service they already know is building on years of context. GCC leaders can support that learning by giving teams access to domain specialists and customer feedback, along with the authority to question requirements. Judgement grows when people are trusted to use it.
Education should give students practice in making and defending decisions.
Team projects can introduce shared interfaces and changing requirements. Design reviews can ask students to explain alternatives and defend their choices. When AI assists development, students should validate its output, modify it and explain the result.
A student team working with synthetic transaction data could be told midway that legitimate customers in one market are being blocked, then asked to find the cause and explain the trade-off in their fix. Strong foundations in databases, statistics, networks and software design make that reasoning possible. Employers then add the domain depth and supervised responsibility through which expertise develops.
We have to hire for how candidates reach decisions.
Give candidates an unfamiliar problem, a piece of AI-assisted code or a requirement that changes midway. Ask what they would investigate, which assumptions they would test and how they would explain the trade-offs. Assess the reasoning behind the answer and how it develops when new information arrives.
As more global products are built and led from India, selection and development must reflect the decisions those teams will own. That starts with what we choose to assess in the next interview.
Authored by Nishant Chandra, Co-founder – Newton School
