What Is a Forward Deployed Engineer? A Practical Guide

I work as a forward deployed engineer, and I get asked what that means about once a week. This is the answer I give, plus the part most explainers skip: when you should hire one, and when you should not.

Where the role came from

The title was popularised by Palantir, where forward deployed software engineers are internally called "Deltas." Palantir describes them as engineers who deploy its software platforms to customers and whose mandate is to achieve technical outcomes for those customers. One Palantir engineer who moved into the role describes Deltas joining the first customer conversations to explore, identify and scope the customer's business problems.

That is still the core of it. What changed is that AI made the job relevant to almost every software company. A large language model is a capability, not a solution. Somebody has to sit with an operations team, learn how invoices actually arrive or how support tickets actually get triaged, and turn the model into a working system with integrations, permissions and tests. That person is an FDE.

Why demand is rising

The hiring data moved fast. Indeed job postings for forward deployed engineers in the US went from 643 in April 2025 to 5,330 in April 2026, a rise of about 729%, according to Indeed data reported by Business Insider. In India, FDE postings on Indeed grew nearly 19-fold between June 2025 and June 2026.

Two cautions on those numbers. They count postings, not hires, and they start from a small base, so percentages look dramatic. But the direction lines up with what enterprises report about AI more broadly. McKinsey's 2025 survey found that 62% of organisations are at least experimenting with AI agents, but only 23% are scaling them anywhere in the enterprise. The gap between experimenting and scaling is exactly the gap an FDE is hired to close.

What a forward deployed engineer actually does

The job description varies by company, but the work falls into four stages.

1. Find the real problem

The request that arrives is rarely the problem. "We want a chatbot" often means "our support team answers the same thirty questions and cannot find the right document fast enough." An FDE spends the first days with the people who do the work, reads the tickets or the spreadsheets, and writes down what success would look like in their terms.

2. Scope something shippable

Next comes a narrow first version that can go to production, not a demo. That means choosing which workflow to automate first, which systems it must read from and write to, and which actions need a human to approve them.

3. Build it end to end

FDEs are full-stack by necessity. In a single week I might write a document-extraction pipeline, design a database schema, build the review screen an operations person will use, and wire up the API to a CRM. At Avenuor, where I am CTO and founding engineer, I built the platform from zero to production, including AI document processing that turns a photographed invoice into structured data validated against a schema before anything is saved, a RAG assistant and agentic workflows. None of those pieces would be useful without the others.

4. Hand over a system, not a prototype

The last stage is the one that separates an FDE from a hackathon. The system needs logging, evals, access control, a runbook and someone on the customer side who understands it. If it only works while the FDE is in the room, the job is not done.

FDE vs solutions engineer vs consultant

These roles overlap, and the titles are used loosely. This is how I separate them.

The short version: a consultant tells you what to build, a solutions engineer shows you what could be built, and an FDE builds it in your environment and stays until it works.

When you should hire a forward deployed engineer

An FDE is the right hire when three things are true:

• The problem is specific to your operations. Off-the-shelf AI tools solve generic problems well. If your process involves your own documents, your own approval rules and three internal systems, someone has to build the glue.
• You need production, not a pilot. If the goal is a slide that says "we tried AI," you do not need an FDE. If the goal is a workflow your team uses every day, you do.
• You lack someone who can span the stack. Many teams have strong backend or data people but nobody who can talk to operations staff in the morning and ship a frontend, an agent loop and an integration by the afternoon.

You probably do not need one if a configured SaaS product already solves the problem, or if you have an internal platform team with spare capacity and AI experience.

Full-time FDE, contractor, or agency?

Most companies hiring the role full-time are AI vendors and platforms placing engineers with their own customers. If you are a business adopting AI rather than selling it, a full-time FDE only makes sense once you have a pipeline of several AI projects. Before that, an independent FDE engaged per project tends to fit better: you get the same person from discovery to handover, without building a team you might not need in a year. I compared the trade-offs in more depth in freelance AI engineer vs agency, and the cost side in how much it costs to hire an AI developer.

What to look for when you hire one

Whichever route you choose, these are the signals I would check:

• Production evidence. Ask what they shipped that real users rely on today, and what broke after launch.
• Discovery questions. A good FDE asks more about your process than about your tech stack in the first call.
• Full-stack range. They should be comfortable with data, backend, frontend and the model layer, because the work will touch all four.
• A view on evals and guardrails. Ask how they would know the system is getting worse. If the answer is "users will tell us," keep looking. I wrote about this in why AI agent projects fail.
• Communication across time zones. Much FDE work is remote now. Written specs, recorded walkthroughs and clear handover documents matter more than overlapping hours.

How I work as an FDE

I have spent four years building AI and full-stack systems for clients in the US, the UK, Canada, Germany, the UAE, Australia and India, with a Top Rated Plus badge and 100% Job Success on Upwork. Before Avenuor I was a Frontend AI Engineer at Mediflow 360, building interfaces for a healthcare and legal-tech platform where speed and data sensitivity both mattered. The pattern on every engagement is the same: a discovery call, a written scope with one shippable first version, weekly demos of working software, and a handover your team can run without me.

Work with me

If you have a workflow that AI should be handling and nobody to own it from discovery to production, I can help. See how I approach AI agent development and workflow automation, look at what I have built, or book a scoping call and we will work out whether a forward deployed engagement fits.