What is a forward deployed engineer?

What a forward deployed engineer does

An FDE sits between a product and one customer's messy reality. PostHog's explainer describes the job as closing the gap between what the product does and what the customer needs, whether that means working from the customer's office, joining their Slack or getting access to their infrastructure (PostHog, February 2026).

The work is mostly engineering. OpenAI's posting for an FDE in Zurich lists the job as owning "technical delivery across multiple deployments from first prototype to stable production", embedding with customer teams, and sharing "field feedback that helps Research and Product understand where the models succeed and where they can improve" (OpenAI posting).

A product engineer builds one feature for thousands of users. An FDE builds what one customer needs this month, then tells the product team which parts other customers will want too.

Where the role came from

Palantir created it. Showing intelligence software to government customers could take weeks of NDAs and security clearances, so Palantir sent engineers to work on site, "forward deployed" with the customer, according to PostHog's history of the role.

AI companies picked the model up because their customers have the same problem in a new form. As Anjor Kanekar, an FDE at Palantir, told PostHog: "AI companies have the models, but they need to figure out what exactly needs to be built." PostHog also cites a Financial Times report of 800% growth in FDE job postings during 2025, and links a16z calling it "the hottest job in startups".

FDE vs software engineer

The two roles use the same skills in different proportions. PostHog's comparison comes down to who you build for and how finished the work needs to be.

At some companies the title covers work close to sales engineering. PostHog notes that some FDE roles are "much closer to technical consultants or sales engineers", so read the responsibilities in a posting rather than trusting the title.

Can a fresher become a forward deployed engineer in India?

At some companies, yes. Sarvam AI's posting for a Forward Deployed Software Engineer in Bengaluru asks for 1 to 3 years of experience and lists "Previous customer-facing experience (including internships)" among its preferred qualifications (Sarvam AI careers).

That role builds LLM applications on Sarvam's APIs, uses prompt engineering, supports pre-sales demos and relays customer feedback to the product team. It asks for Python, REST APIs, and an understanding of vector databases and semantic search.

Compare that with OpenAI's Zurich posting, which asks for "5+ years of engineering or technical deployment experience that includes customer-facing work". For most students, the realistic path is a software or AI engineering role first, with customer-facing projects or an internship that proves you can work with the people who'll use your code.

Skills to build as a student

Build the skills both postings share, in this order. Each one should end in a project you can demo in five minutes.

1. Write a Python integration that copies records between two systems over their REST APIs and survives pagination, rate limits and failed requests.
2. Join two messy datasets in SQL, such as a customer list and a payments export, and explain every row you dropped.
3. Build an LLM retrieval app over real documents, with a small test set of questions and the answers you expect.
4. Package one project with Docker, keep secrets out of the code, and add logs you can read when it breaks.
5. Write a one-page scope for each project and record a short demo aimed at someone non-technical.

The first item has the most edge cases. Here is the shape of a fetch that handles pages and rate limits in Python:

import time
import requests

def fetch_all(url, token):
    """Fetch every page from a paginated API, backing off on HTTP 429."""
    items, page = [], 1
    while True:
        for attempt in range(5):
            r = requests.get(url, params={"page": page},
                             headers={"Authorization": f"Bearer {token}"}, timeout=10)
            if r.status_code != 429:
                break
            time.sleep(2 ** attempt)
        r.raise_for_status()
        batch = r.json()["data"]
        if not batch:
            return items
        items.extend(batch)
        page += 1

In an interview, the code matters less than the reasons. Be ready to say why you back off exponentially, what happens after the fifth attempt, and how you'd resume if the job died on page 40.

How to get customer-facing experience as a student

Sarvam AI's posting counts internships as customer-facing experience, so find a real user before you graduate. The project matters less than the fact that someone else depended on it.

Offer to build something for a college club, a department office or a family business. A sign-up sheet that syncs into a spreadsheet, or a dashboard of the club's event attendance, is enough if real people use it and tell you what's wrong with it.

Keep a log of every request, what you built in response and what changed after people used it. In an FDE interview, that log is your evidence that you can turn a vague request into working software. If you take a remote internship to get this experience, our guide on whether virtual internships are worth it covers how to make one count.

How the DevIt FDE internship maps to this

DevIt's Forward Deployed Engineer internship is built around the same loop of discovery, prototype, deployment and support. It expects basic Python, a terminal and Git before you start.

The 1-month plan includes a two-system data sync (DataBridge), a customer metrics dashboard (InsightDash) and a customer onboarding tool as the capstone. The 3-month plan adds a third-party API connector, a domain data model served over an API, containerised multi-environment delivery, an observability and support runbook project, and a deployed customer-facing capstone.

It won't make you an FDE by itself. It gives you graded projects that match what these postings ask for, plus a certificate with a verifiable ID. It's a paid training internship with no stipend: on 12 October 2026, the 1-month plan cost ₹299 at its Early Bird price.

If the LLM side interests you more than integrations, the Generative AI internship covers embeddings, semantic search and retrieval. Every track is listed on the internships page.

Job postings and DevIt track details checked 12 October 2026. Postings close and change often, so check the company's careers page.