Indian IT firms race to build forward-deployed AI teams

Indian technology services companies are assembling large teams of forward-deployed engineers as customers seek faster, more practical applications of artificial intelligence. The Times of India reports that Infosys, Tata Consultancy Services and LTIMindtree are among the companies building or expanding capabilities for the role, known as FDE.

An FDE works directly with a client to identify a useful problem, develop a solution and put it into operation. The role combines software engineering with customer judgment and responsibility for the eventual result. That makes it different from a conventional services assignment, in which engineers typically implement requirements defined by someone else.

Infosys plans to expand its frontier engineering group to 6,000 engineers, while TCS is reportedly creating an FDE pool of 8,900 people. LTIMindtree has announced a programme to produce more than 1,000 AI-certified engineers, including people prepared for forward-deployed work.

The rapid expansion reflects a wider shift in the technology industry. AI is reducing some traditional software tasks while increasing demand for engineers who can apply new tools to complicated business settings. The result is a shortage of people with both advanced technical ability and the confidence to work closely with customers.

A role that combines rare skills

The FDE approach was pioneered by US software company Palantir. Its engineers work alongside clients, determine what needs to be built and help deploy it, rather than stopping when a technical specification has been completed. Industry executives say Indian companies cannot reproduce that model simply by moving more employees into client locations or changing the title of existing service workers.

Anjor Kanekar, a former Palantir FDE and founder of Platypus Technologies, told The Times of India that the job demands three capabilities at once: production-level engineering, the judgment to select a worthwhile business problem and the composure to make decisions in front of a customer. Those qualities are not commonly found together. Many accomplished engineers prefer to remain away from customer-facing work, while many effective client-facing professionals do not build production software.

Vishal Sikka, former Infosys chief executive and founder of Hangten Systems, said exceptional problem-solving would not grow in direct proportion to headcount. In his view, simply redesignating onsite employees as FDEs while retaining hourly billing would fail to capture the purpose of the model. He argued that AI is weakening the traditional connection between revenue and the number of people assigned to a project, creating an opportunity for India to move beyond supplying labour and towards increasing the impact of skilled workers.

Viral Shah, chief executive of JuliaHub, pointed to Palantir’s organisational design as another obstacle to direct replication. Palantir generally uses small project teams staffed by young engineers familiar with modern development tools and the company’s own products. Those teams also have substantial freedom over technology choices, team composition and hiring, allowing them to move quickly.

Why conventional training may fall short

Indian IT companies have adapted to earlier technology changes, Shah said, but the AI transition is different because it challenges the labour-arbitrage model on which much of the industry developed. He added that companies could adapt if their leaders are technically engaged and understand how AI-native teams operate.

The supply of potential FDEs is especially difficult to expand because broad AI education does not automatically produce the required specialists. Cognizant chief learning officer Thirumala Arohi said the role depends on technical depth, ownership of a business domain and a distinctive approach to solving unfamiliar problems. Those abilities cannot be reliably produced by placing courses on a training schedule.

Cognizant has given 350,000 associates AI-fluency training, with more than 150,000 included in its AI Bridge programmes. Arohi said such efforts raise general capability across the organisation, but only a smaller group will be able to enter a claims operation or another client environment and take responsibility for its result. The gap, he said, is an intentional feature of the selection process rather than evidence that mass training has failed.

Scaling through small teams

The emerging answer may be to reproduce compact, autonomous units instead of continually enlarging individual FDE teams. Cognizant uses pods of one to three people, with the size deliberately limited to prevent handoffs from returning between problem definition, construction and accountability. Growth comes from adding more pods, not from expanding each pod.

Lalit Wadhwa, chief technology officer at Coforge, said an effective FDE must be able to discover and define a problem, build a production-ready answer, support its adoption and assess the outcome. Developing those abilities requires sound engineering judgment, authority to make appropriate decisions and regular feedback from platform teams.

Ramesh Lokre, co-founder and chief executive of Saicon Consultants, said the central issue was broader than recruitment or instruction. Forward deployment requires a sense of ownership and an enterprise-wide outlook. If companies merely rename the old onsite-offshore arrangement, he warned, they will not create the model customers are now seeking.