FDE job postings in India grew almost 19-fold: where demand sits and what three listings ask for

Almost 19 times. According to Indeed data cited by People Matters, that is how much job postings for Forward Deployed Engineers in India grew in 2026 compared with the same period a year earlier. Postings peaked in June, and by August they were still well above the level of a year before.

Supply is far thinner. Research by CIEL HR estimates that India has only about 1,200 people able to take on the full role, and about 6,000 if those with adjacent skills are included.

With demand soaring and few people ready, the question worth asking is whom companies are choosing and on what evidence. A close reading of open listings shows a market short of people but not lowering its standards across the board: there is a door for those with few years of experience, while senior roles demand very specific things.

Demand clusters in three cities

According to Indeed, FDE postings in India fall into six location groups: Bengaluru, Hyderabad, Chennai, New Delhi, Mumbai and remote. The list looks fairly scattered, but CIEL HR’s figures show demand is much more concentrated. Bengaluru accounts for almost half of annual FDE demand, followed by Delhi-NCR and Hyderabad, and together the three make up almost 80% of postings.

Individual listings match that picture. NTT DATA is hiring FDEs in Bangalore. GlobalLogic has opened a hybrid role across Bangalore, Hyderabad and Noida, and Noida is part of the Delhi-NCR region. The listing posted by HuntingCube is based in Bengaluru East.

If you are outside India and eyeing the remote group, read each listing carefully before applying. A listing marked remote may simply mean no office attendance, not that it accepts candidates from any country. The quickest route is to ask the recruiter directly about residency and contract type in your very first message.

Three listings, three thresholds

According to the employers quoted by People Matters, they need engineers who can sit with customers, clarify the real problem and make technology work in practice. Open roles span several areas: enterprise AI, data engineering, process automation, infrastructure and healthcare.

But when it comes to writing the listing, each company sets the bar in a different place.

	NTT DATA	GlobalLogic	HuntingCube
Location	Bangalore	Hybrid: Bangalore, Hyderabad, Noida	Bengaluru East
Experience	Typically 8 years or more	5-10 years	1-3 years, including internships at product companies
AI technology	GenAI, agents, RAG	Google Cloud, Vertex AI	Projects using LLM APIs count, including personal ones
Listing’s emphasis	Enterprise customers, leading discovery and prototyping; evals, security, observability	Automated eval loops measuring performance, drift, accuracy; a technical voice in front of customers	Internships and side projects count towards years of experience

NTT DATA’s listing leans heavily towards consulting. It wants people already used to working with enterprise customers who can lead on their own from discovery through to a prototype. GlobalLogic’s listing leans towards the platform, tied to a specific stack: Google Cloud and Vertex AI.

HuntingCube’s listing is the most notable exception. It asks for only 1-3 years as a software engineer, internships at product companies count, and a project using an LLM API, even one built on the side outside working hours, is treated as experience. For people early in their careers, that is a real opening.

At senior level, customers and evals are mandatory

The two senior listings converge on the same point. NTT DATA requires applied knowledge of generative AI, agent systems and RAG, along with the concerns that come with running systems in production, such as evals, security and observability. GlobalLogic writes directly into the job description that the FDE must build automated eval loops to measure model performance, drift and accuracy.

Why do evals appear in the job description of a customer-facing role? Because they are how an FDE proves to the customer that the system is working correctly, and catches it in time when it starts to go wrong.

Imagine you deploy a RAG assistant that answers questions about internal procedures for a hospital. You write 50 questions with reference answers, rerun them every time you change the prompt or the model, and record how many answers are correct.

If the first run gets 41 right and after a prompt change only 37, you know at once that you have just made the system worse, before any doctor notices.

Read the three listings in order of years required and a path emerges. At 1-3 years, a project that calls an LLM API already counts as experience; to move up to the 5-10 year level, the listings state plainly that you must know how to measure, and how to stand in front of a customer and defend those numbers.

Will a shortage widen the door?

Not necessarily. CIEL HR says demand is strongest for those with 5-8 years of experience, and both the NTT DATA and GlobalLogic listings fall within that range or above it. The pay gap between levels is also wide: early-stage FDE roles pay ₹35-45 lakh a year, while seasoned specialists earn ₹70-90 lakh.

With the pool of ready candidates so thin, one sensible route for companies is to train FDEs from their existing staff. Remarks by the CEO of TCS suggest at least one large company is thinking along these lines.

He said the company wants as much as 1%-1.5% of its workforce to be people who could be called FDEs, equivalent to roughly 5,900 to 8,900 people by Business Today’s estimate. The article also noted it is unclear whether this means new hires or retraining existing staff.

Either way, the signal for engineers at large services companies is clear: the route to becoming an FDE may run through an internal transfer rather than a job change. If your company is looking for people for this role, put yourself forward early and ask to join projects with direct customer contact.

Turning listing requirements into a portfolio

This market rewards evidence over titles. If you have 1-3 years of experience, build a project with an LLM API that solves a problem for real users, with an eval set and a README that records the numbers before and after each improvement.

The LLM API part matches what HuntingCube’s listing says will count; the eval part is early preparation for the requirement GlobalLogic spells out at the 5-10 year level.

If you are already at 5-8 years, the gap is usually not technical but on the customer side. Rewrite your CV so that each project tells three things: what problem you clarified with the customer, what prototype you built, and how you measured the result.

The phrase “discovery and prototyping” from NTT DATA’s listing should appear on your CV, backed by a real example.

When reading a senior listing, look for two signs: does it say clearly whom you will work with on the customer side, and does it mention evals or observability? A senior listing missing both may just be an old engineering role with a fashionable title attached.

With 1,200 ready people facing a market that has grown almost 19-fold, the vacancies are likely to stay open for some time. Those who fill them will not necessarily be the people with the most years of experience, but the ones who can produce an eval number and tell the story of sitting down with a customer.