Front-end
Back-end
Full Stack Developer
Mobile
Other
Company
Contact Us
A forward deployed engineer (FDE) is a software engineer who works inside a customer's team and systems to find the real problem, build the fix and deploy it, then stays until people use it. The role started at Palantir and is now one of the fastest growing jobs in software. US postings rose more than 1,000% in the first eight months of 2026 compared with the same months of 2025 (Fortune, using Lightcast data).
This guide is for the people deciding whether to hire one: founders, CTOs and the leads who own an AI rollout. It covers what the job involves, how it differs from similar roles, what it pays, and how to tell whether your company needs one.
The short answer
- What it is: a software engineer who builds and deploys inside the customer's environment, not from behind a ticket queue.
- What it pays: a median of $195,000 across 924 US job posts in 2026 (Recruiting from Scratch).
- How it differs from a solutions engineer: a solutions engineer shows that the product fits. A forward deployed engineer builds what the customer needs, including things the product does not do yet.
- When you need one: your AI works in a demo, but it has not reached daily use inside your systems.
What does a forward deployed engineer do?
A forward deployed engineer owns the gap between a working product and a working result. Palantir calls these engineers "Deltas" and pairs them with deployment strategists called "Echos" (Palantir). The Delta writes the code, and the Echo manages the relationship and the workflow.
A typical engagement runs through four kinds of work, often at the same time:
- Discovery. The engineer sits in on the work itself: support calls, finance close, a claims queue. The goal is to write down the real problem, which is rarely the one in the original request.
- Building. Connections to the systems the customer already runs, such as a CRM, an ERP, a data warehouse or an identity provider. Sometimes it means building the missing feature outright.
- Deployment. Access approvals, audit logs, security review answers, and tests on the customer's own data before anything goes live.
- Adoption. Training the team, watching real usage, fixing what people actually hit, and handing over a runbook.
The last step is what separates the role from ordinary contract development. A forward deployed engineer is judged on whether the customer uses the system, not on whether the ticket closed.
What comes back from the field matters too. Good forward deployed engineers feed what they learn into the product team, so the next customer needs less custom work.
Where did the role come from, and why is everyone hiring now?
Palantir created the title in 2011 by renaming its solutions and integration engineers (a16z). The new name made client-facing engineering a role strong engineers wanted. For years it stayed mostly a Palantir term.
AI changed that. A model is easy to demo and hard to deploy, because the hard parts sit around the model: data access, permissions, tests on real work and people changing how they work. Andreessen Horowitz partner Joe Schmidt argued in June 2025 that complex AI products need this kind of hands-on implementation to succeed. He also noted that 22 of OpenAI's 311 open roles at the time were forward deployed or solutions engineering positions (a16z).
By 2026 the role had spread well beyond AI labs. Microsoft, Meta, Google, OpenAI and Anthropic have all opened forward deployed engineering positions. Lightcast data shows postings up more than 4,600% compared with 2023 (Fortune).
So why does this matter if you are not an AI lab? Because your vendors are hiring the same people you would want. Competing with them for full-time hires is slow and expensive, which is why the hiring options later in this guide matter.
What we see in our own work: Our engineers also write the tasks used to test frontier AI models for AI programmes. The tasks a model fails most often are the ones where a professional relies on context nobody wrote down. Much of a forward deployed engineer's job is finding that context and turning it into checks before the system goes live. You can see a scored example on our AI training data page.
Forward deployed engineer vs solutions engineer vs software engineer
The titles overlap, and companies use them loosely. The clearest way to separate them is to ask what each person is measured on.
| Role | Main job | Writes production code | Time with customers | Measured on |
|---|---|---|---|---|
| Forward deployed engineer | Builds and deploys what the customer needs, inside the customer's systems | Yes, daily | Daily, as part of the customer's team | The result running in production and in use |
| Solutions engineer | Shows that the product fits the customer's needs | Demos and proofs of concept | Mostly before the sale | Deals won |
| Product software engineer | Builds the core product for every customer | Yes, daily | Rarely | The product roadmap |
| Implementation consultant | Configures the product for one customer | Some scripting and configuration | During the rollout | Project delivered on time |
You'll also see these titles for the same job:
- Forward deployed software engineer (FDSE): Palantir's longer form of the title.
- Forward deployed AI engineer: used by AI companies for LLM and agent deployments.
- Applied AI engineer and deployment engineer: common at AI startups.
The name matters less than the scope. If the person writes production code, works directly with the people who will use it, and stays until it is in daily use, the role is forward deployed.
How much does a forward deployed engineer earn?
The median US salary for a forward deployed engineer is $195,000, based on 924 job posts analysed by Recruiting from Scratch in June 2026. The middle half of roles pay between $160,000 and $215,000 (Recruiting from Scratch).
Other data points line up:
- Against ordinary software roles: Lightcast puts the median advertised salary above $188,000, against about $145,000 for a traditional software engineering role (Fortune).
- By location: the San Francisco median is $194,000 and the remote median is $180,000 (Recruiting from Scratch). Remote forward deployed roles are common, which surprises people who assume the job means sitting at the customer's desk.
- At the top of the market: AI labs pay well above the median, so treat headline offers as outliers rather than the going rate.
What it costs the employer is higher than the salary. Add payroll taxes, benefits and equipment, which our software development cost benchmark puts at about 35% for a US hire. On a $195,000 salary that comes to roughly $263,000 a year, or about $21,900 a month, before recruiting fees.
When does your company need a forward deployed engineer?
You need one when the work sits between systems and teams, and nobody owns getting it into daily use. Five signs come up again and again:
- A pilot that never shipped. The demo worked months ago and it still isn't part of anyone's day.
- Customers waiting on integration. If you sell an AI product, enterprise buyers want help connecting it before they sign.
- Several systems in the path. The workflow touches a CRM, a finance system, a data warehouse and an approval step, and each has its own owner.
- Requirements that only make sense on site, or on the call. The real rules come out when someone watches the work being done.
- A live system nobody uses. It passed every test, and the team went back to the spreadsheet.
You probably don't need one if the work is a well-defined feature inside your own product. A dedicated developer fits that better and costs less. The same goes for configuring a SaaS tool exactly as its vendor designed it, which the vendor's own implementation team can usually do.
How do you hire a forward deployed engineer?
There are three routes, and they differ most on speed and on who carries the risk.
| Route | Time to start | Cost | Best when |
|---|---|---|---|
| Full-time hire | Months, against strong competition for the same people | Salary near $195,000 plus employer costs | The role is permanent and central to your product |
| Contractor through a talent marketplace | Days to weeks | Hourly, varies widely, plus platform fees | A short, well-scoped job |
| Remote forward deployed engineer from a development company | About a week | A fixed price per task, hourly or monthly | You need the work done now and may not need a permanent hire |
Whichever route you take, look for these in the person:
- They have shipped into systems like yours. Ask for a deployment that went wrong after launch, and what they changed.
- They test on real work first. Ask how they would check the system on your data before anyone relies on it.
- They write things down. You want a plan, decisions and a runbook, not knowledge that leaves with the person.
- They are comfortable on the call. Much of the job is talking to the people who will use the system, and to the ones who don't want it.
Empiric Infotech provides remote forward deployed engineers. The engineer works inside your Slack, repositories and calls, joins meetings outside regular development hours when your calendar needs it, and owns the rollout from discovery to adoption. Most work is bought as a fixed price per task; you can also pay USD 70 an hour or USD 10,000 a month for a full-time engineer, with regional rates on the page. See forward deployed engineers for how it works.
Frequently asked questions
What does FDE stand for?
FDE stands for forward deployed engineer. Palantir's longer form is forward deployed software engineer, or FDSE. AI companies also use forward deployed AI engineer for the same job. In every version, "forward deployed" means the engineer works inside the customer's team and systems rather than only on the vendor's product.
Is a forward deployed engineer the same as a solutions engineer?
No. A solutions engineer mostly works before the sale, showing that the product fits. A forward deployed engineer works after it, writing production code inside the customer's systems and staying until the result is in daily use. Some companies blur the two titles, so check the job's scope rather than its name.
Do forward deployed engineers work on site?
Some do, especially in government and defence work, where Palantir's model began. Many now work remotely. Recruiting from Scratch's 2026 data shows a remote median salary of $180,000, which tells you remote roles are common. The work happens in the customer's systems, calls and chat, wherever the engineer sits.
How do I become a forward deployed engineer?
Most come from full-stack or data engineering backgrounds. They add integration work, customer conversations and comfort with unclear requirements. Shipping an integration end to end and talking to its users is the best practice. If you are looking for open roles at Empiric Infotech, see our careers page.
