What is a forward deployed engineer?

Last updated: 29 October 2026

A forward deployed engineer (FDE) is a software engineer who works inside a customer's organisation, on its systems and data, until software is running in production. Palantir pioneered the role. In 2026 AWS (30 June, $1 billion) and Microsoft (2 July, $2.5 billion) launched their own forward deployed engineering units.

FDE stands for forward deployed engineer. Palantir's own title is forward deployed software engineer (FDSE). The engineer doesn't advise from outside or build from a distance. They sit where the work happens, learn how it is done and build the thing that changes it.

The term is now attached to a lot of AI work, and it is used loosely. This guide covers where the role came from, what an FDE does, how the role differs from a software engineer, a solutions engineer, a consultant and an outsourced team, and when a company needs one.

Where did the forward deployed engineer role come from?

The role is usually traced to Palantir. Its job posting says: "We pioneered this unique position, embedding talented engineers directly with our customers". The same posting calls the role "the blueprint" (read in 2026).

The idea spread with AI. Large models are easy to try and hard to fit into a real business process. Vendors found that the gap between a promising trial and software people use every day was mostly engineering work inside the customer. So they started sending engineers in.

2026 made it official at the largest vendors:

• AWS, 30 June 2026. Amazon launched a forward deployed engineering organisation and committed $1 billion to it, from internal resources. Its engineers "will embed within companies to deploy purpose-built agents", TechCrunch reported. The same report notes that OpenAI and Anthropic had launched their own FDE joint ventures in the months before.
• Microsoft, 2 July 2026. Microsoft announced the Microsoft Frontier Company, its move into forward deployed engineering, with 6,000 engineers and salespeople and a $2.5 billion budget, Directions on Microsoft reported. The article notes it isn't clear whether the budget is new money.

Directions on Microsoft also notes that the Amazon, Anthropic and OpenAI offers focus on the adoption of their own AI products. That point matters later, when we look at the forward deployed engineering model from the buyer's side.

What does a forward deployed engineer do?

Palantir's description is a good summary. Its engineers "work directly with customers to quickly understand their greatest problems", and a day "might span discussing architecture with fellow engineers, wrangling massive-scale data" (job posting, read in 2026).

In practice, the work has three parts.

1. Learn how the work is done today. The engineer sits with the people who do it. If the problem is invoices, they match invoices by hand with the finance team until they know the rules, including the ones nobody wrote down.
2. Build it inside the customer's systems. The code lives in the customer's cloud and repositories, under the customer's access controls. The engineer works with the data the company already holds, not a clean sample.
3. Hand it over. The job ends when the customer's own team can run what was built. That means documentation, training and a handover someone signs off.

The engineer is a builder first. They write production code. They also talk to the head of operations and the IT manager, because the hard part of most AI projects isn't the model. It's the supplier names that differ between two systems, or the approval step nobody mentioned.

What skills does a forward deployed engineer need?

The role mixes two kinds of skill:

• Engineering. Writing production code, working with messy data and connecting systems through their APIs.
• Working with people. Interviewing staff about their work, explaining trade-offs to managers and saying no to features that don't serve the goal.

Many engineers have one half. The role needs both, which is why it is hard to hire for.

Forward deployed engineer vs software engineer, solutions engineer, consultant and outsourcing

These roles get confused because they overlap. The difference is what each one is for, and who answers for the result.

In one line each:

• A software engineer builds your product and stays.
• A solutions engineer shows how a product could fit, then hands the build to someone else.
• A consultant tells you what to do.
• An outsourced team does what the tickets say.
• A forward deployed engineer works out what to build by watching the work, builds it in your systems and leaves your people able to run it.

None of these is wrong. If you need a strategy, hire a consultant. If you have a clear spec and someone to manage it, outsourcing works. If the work is your core product, hire software engineers. The FDE model fits the space in between: you know where the manual work is, but not yet what to build or how it fits your systems.

When does a company need a forward deployed engineer?

Signs it fits:

• You have tried AI and it stalled. A pilot worked in a demo and never reached the people who do the work.
• The problem lives in a process, not a product. Documents to read and check, requests to route, data moved between two systems by hand.
• Your team can't spare the time. Your engineers are busy with the core product, and nobody owns the AI work.
• The knowledge is in people's heads. The rules for the process exist only in how your staff do it today.

Signs it doesn't:

• You need a new product built from a fixed spec. A product team or a fixed-price build fits better.
• You want a permanent hire. A forward deployed engineer hands over and leaves; a hire stays.
• Nobody inside can give the engineer access, data and an hour of their time each week. The model depends on that.

What is the forward deployed engineering model?

The forward deployed engineering model is a way of delivering software where engineers work inside the customer's business until the result runs in production, instead of building elsewhere and shipping it in. You can get it in two ways.

From a platform vendor. Vendor FDE teams, such as those at Amazon, Anthropic and OpenAI, focus on getting customers to adopt the vendor's own AI products. That is useful if you have already chosen the vendor. The work centres on their technology.

From an independent team. Sometimes called forward deployed engineer as a service. You get the same way of working, without committing to one vendor's platform, and the work is aimed at your use case rather than at a product sale.

This is also what people mean by an AI implementation partner: an outside team that takes an AI use case from idea to production inside your business, and answers for the result.

How iHux offers a forward deployed engineer

iHux offers a forward deployed engineer to companies. We also call it an embedded engineer: the two names mean the same offer. A senior engineer works alongside your people, on your systems and data, with our team building behind them, until an AI use case is running in production. Then we hand it over.

It works in three stages:

1. Discovery, two weeks. The engineer sits with your team, maps how the work is done today and returns a short list of AI use cases ranked by value and difficulty. You keep the report. If you continue, the discovery fee is credited to the first month.
2. Embed, billed monthly. One use case at a time, taken to production, with a working demo every two weeks. A second senior engineer reviews every change before it reaches your systems.
3. Handover. Documentation, training for the people who will run it, and an optional maintenance plan.

The facts in short:

• Two-week discovery first, then a quote.
• One engineer embedded, one use case at a time.
• A second senior engineer reviews every change.
• Pause or end with 30 days' notice. There is no published minimum term.
• The code, models, prompts and documentation are yours from the first commit, in your accounts.
• Quote only: every engagement is priced for your team, so there is no price list.

To start with a two-week discovery, talk to us about a forward deployed engineer. If your first use case involves documents, read AI document processing: what it can and can't do. For why AI pilots stall before production, read why AI projects fail and how to reach production. Our AI-powered product development service covers building AI products from scratch.