What Is a Forward Deployed Engineer? Role, Skills & Salary

A forward deployed engineer (FDE) is a software engineer who embeds with a customer’s team to deploy, customize, and integrate a company’s product inside the customer’s actual environment. Palantir invented the title around 2009. AI labs, AI startups, and enterprise software companies have been copying it since roughly 2023, and the job descriptions under that banner now vary so widely that two FDE offers at two different companies can mean completely different careers, different comp philosophies, and different definitions of what a good Tuesday looks like.

That confusion is the main reason this post exists. If you’re a hiring manager writing an FDE req, or a candidate staring at an offer letter and wondering whether you’re really being hired as an engineer or as a glorified consultant with a git commit history, the role’s definition matters before the comp does.

Mike Carter, KORE1. I handle AI and software engineering searches for our IT staffing practice. FDE reqs have been landing on my desk every few weeks since the middle of 2024, and the pattern is consistent enough that it’s worth writing down. Bias note: we get paid when clients hire through us, and part of why I’m writing this is so the next hiring manager who opens an FDE req calls us. The rest of why I’m writing it is that I’ve watched three separate clients hire the wrong kind of FDE and burn four to six months figuring it out. Expensive lesson. Avoidable.

What a Forward Deployed Engineer Actually Does

Short definition for snippet eligibility: a forward deployed engineer is a customer-embedded software engineer who deploys, integrates, and customizes a vendor’s platform inside a specific client’s infrastructure, then writes code against the real data the client sees in production. The job blends full-stack engineering, solutions architecture, and the parts of product management nobody wants to call product management.

The title comes from Palantir, where FDEs were the people who actually made Foundry and Gotham work for intelligence agencies and hospital networks after the sales team left the building. Palantir’s own description still reads like a mission statement about shaping the world’s most important institutions. AI-era FDE roles at places like OpenAI, Anthropic, and a long list of Series A and Series B startups borrow the title, but the job on the ground looks different at each one depending on how the company thinks about the split between product engineering and customer-embedded work.

What stays consistent across employers:

• The engineer is physically or virtually embedded at the customer, not sitting at HQ
• Production code gets written against the customer’s data, not a toy dataset
• Shipping cadence is measured in weeks or customer milestones, not quarterly release trains
• The engineer is the single technical point of contact for that customer
• Success looks like the customer actually using the thing, not just paying for it

What varies wildly: how much of the job is new product development versus customization, how much sales involvement there is, how many customers a single FDE covers, and whether travel is a requirement or a rumor.

The Three Types of FDE You’re Actually Hiring For

Before you write the job description, figure out which of these three versions you mean. Getting it wrong costs months.

We placed an enterprise integration FDE at a GenAI startup last summer, and the story is a clean example of how misreading the flavor torches a search. The client had initially written the req looking for the third flavor, a product FDE, because that’s what the founder had done at their previous company. The actual need was the first flavor because they had just closed a global bank as their first enterprise customer and the bank’s security team wanted someone on-prem in a SCIF two days a week. Totally different hire. We rewrote the JD over a 40-minute call, the search closed in 31 days, and the FDE is still there nine months later running point on their second enterprise deployment.

If you cannot tell a hiring manager in one sentence which of those three flavors you need, you are not ready to open the req.

Forward Deployed Engineer vs Solutions Engineer vs Sales Engineer

These titles get conflated constantly, including inside the companies posting them. Here is the actual difference.

Three practical tells. FDEs have a GitHub history that looks like an engineer’s. Solutions engineers have a calendar that looks like a project manager’s. Sales engineers have Salesforce on their desktop.

If a job description uses all three titles interchangeably, it is almost always an FDE role being scoped by someone in sales.

A Day That Looks Nothing Like a Normal Engineering Day

A product engineer writes code, reviews PRs, attends a standup, and goes home. An FDE does all of that plus about four other things that do not fit on a calendar.

Tuesday for an FDE I placed at an AI infrastructure startup last fall, paraphrased from his own description: 7:45 AM local kickoff with the customer’s data platform team in Frankfurt, because that is where the design partner sits. 9 AM pair-coding session over Zoom with a staff engineer at the customer, building a retrieval pipeline on top of the vendor’s API. 11 AM hard pivot because the customer’s security review flagged a dependency. The next two hours are him swapping libraries and writing a Jira ticket back at HQ asking for a feature the customer actually needs. 2 PM demo prep. 3 PM demo with the customer’s director of ML engineering, three product managers he has never met, and someone from procurement who asks about SOC 2. 4 PM a Slack argument with his own product manager about whether this customer’s request belongs on the core roadmap. 6 PM writing a two-paragraph summary of the day for his account executive because the renewal is in six months.

Nothing on that schedule is unreasonable on its own. All of it together is. The FDE hires who last are the ones who can live inside that ambiguity and still ship working code at the end of the week. Most engineers cannot.

The Skills That Actually Matter

Standard job descriptions list Python, cloud platforms, APIs, and “strong communication skills” the way every engineering job description does. That list is not wrong. It is also not helpful, because it describes roughly 40% of the working software engineers in the country. Here is what separates FDE-caliber candidates from the rest.

Shipping speed under ambiguous specs. The customer will never hand an FDE a clean set of requirements. The ask is always some version of “we want to use your thing for this problem” and the FDE has to translate that into a working implementation plan by Thursday, often while three different customer stakeholders are sending conflicting messages about what matters most. Candidates who need three weeks of spec-writing before they touch a keyboard will fail.

Customer-side emotional intelligence. The FDE is sitting in meetings with the customer’s engineers, who often feel threatened by a vendor being that deep inside their stack, and with the customer’s VPs, who want to see weekly progress that the engineers know is unrealistic. Reading that room is a real skill. Most engineers have not been trained for it.

One stack deep, three stacks functional. A good FDE has a native stack they can ship production code in blindfolded (usually Python or TypeScript in 2026), plus working fluency in at least two others so they can integrate with whatever the customer happens to run (Java, Go, Rust, C#, sometimes Scala in banking). You do not need to be an expert in all of them. You need to read an unfamiliar codebase and ship a patch against it inside a week.

Willingness to be wrong in public. The FDE disagrees with their own product team in front of the customer at least once a quarter, and they also disagree with the customer in front of their own sales team at least as often, which means they’re building two slightly different kinds of political capital at once. If a candidate cannot comfortably say “that is the wrong approach and here is why” to a VP on either side of the table, they will quietly nod through bad decisions and the deployment will suffer.

Travel tolerance, if the flavor calls for it. Roughly half of Enterprise Integration FDE roles involve real travel, usually 20 to 40% depending on the customer. Post-Sales AI FDE roles have shifted mostly remote since 2023. Product FDE roles are split about evenly. Ask before accepting.

Forward Deployed Engineer Salary Ranges for 2026

FDE compensation sits at or slightly above equivalent product engineering roles at the same company, because the job is harder to staff and the attrition risk is higher. The following ranges pull from Levels.fyi and Glassdoor data current as of early 2026, cross-checked against our own placement comp at similarly staged companies. Variance between the two aggregators ran about 8 to 15% on any given band, which is normal for a niche role that most salary tools still classify under “software engineer” without any further tagging.

Totals at frontier AI labs run higher, in some cases a lot higher, because equity at a company still private at a $100B+ valuation carries a number on paper that is hard to argue with even when the cap table math looks shaky in the right lighting. We have seen senior FDE packages at the top labs clear $700K all-in. Those offers are not the market. They are their own thing.

For broader comp context on adjacent AI roles, the AI engineer salary guide breaks down comparable numbers for pure ML engineer, applied AI engineer, and research engineer roles. Use the salary benchmark tool if you need a specific city or level.

One caveat worth stating flatly. The FDE role is less than two years old as a widespread title outside Palantir. Salary bands are still stabilizing. A 2026 offer from an early-stage startup might be 30% below a 2026 offer from a later-stage competitor for the same work. Benchmark aggressively before accepting.

Why Hiring FDEs Is Harder Than You Think

Most engineering managers underestimate the FDE hire in two specific ways.

First, the candidate pool is narrower than it looks. You need a strong production engineer who also tolerates meetings, travel, and ambiguity. The Venn diagram of those traits is small. The 2024 Stack Overflow Developer Survey found that roughly 41% of professional developers prefer fully remote work and only 19% are willing to work in-office, which by itself eliminates a meaningful slice of what a real Enterprise Integration FDE search needs. Our AI/ML engineer staffing data shows the subset of AI-capable engineers who will accept a customer-facing travel role sits at around 12% of active candidates in a given quarter.

Second, the interview process most companies run for FDE roles tests for the wrong things. Standard engineering loops are four rounds of LeetCode-adjacent problems plus a system design. That tells you the candidate can pass an engineering interview. It does not tell you whether they will handle a VP of Engineering at a Fortune 100 customer asking them to commit to a three-week timeline during their second on-site. We recommend every FDE loop include at least one mock customer interaction, ideally roleplayed with someone from the vendor’s own field team, and at least one open-ended debugging session against an unfamiliar codebase rather than a clean algorithms problem.

There is a third issue that shows up late in the process. Compensation conversations with FDE candidates are often slower than with product engineering candidates because the candidate is usually weighing two or three other offers from adjacent roles (staff product engineer, applied AI engineer, solutions architect) and the comp bands can be 15 to 20% apart between those titles at the same employer. If your offer is not at least competitive with the alternative titles, do not expect the candidate to take the FDE role out of curiosity. They will not.

For a broader look at how to structure the search, our hiring guide for AI engineers covers adjacent ground in detail, including JD templates and interview structure that ports directly to FDE reqs with minor adjustments. The software engineer staffing page covers the broader engineering search model we use for these roles.

Questions Hiring Managers Actually Ask

Is a forward deployed engineer the same as a sales engineer?

They share a root word and almost nothing else. Sales engineers sit in pre-sales, build POCs, and carry quota with an AE. FDEs sit in engineering, write production code for the customer, and do not carry a quota. The confusion exists because some companies put FDEs under the CRO org for reporting reasons, which blurs everything.

What was the role called before Palantir coined the term?

Variously: solutions architect, technical account manager, professional services engineer, implementation engineer, customer engineer. None of those titles quite captured what Palantir was doing because they included the production code aspect from the start. The new title stuck because it filled a real gap.

How much travel should I expect?

Roughly 20 to 40% for Enterprise Integration FDEs at enterprise-sales companies, sometimes more during deployment phases. Post-Sales AI FDE roles at AI-native companies: mostly remote in 2026, occasional customer on-sites. Product FDE roles during design-partner phases: depends on the partners. Ask during the interview and get a specific answer before signing.

Can I hire an FDE out of a pure engineering background?

Yes, if the candidate has shipped a side project to real users or done tech lead work that involved explaining trade-offs to non-engineers. You need evidence they can navigate ambiguity and interact with humans. A perfect LeetCode record and zero customer-facing experience is a flag.

What’s the career path after FDE?

Three common next steps. Back into product engineering at the same company, usually at a higher level because the FDE work exposed them to more of the stack. Into a staff or principal FDE role if the org has the ladder. Or out, into founder or early-employee roles at startups, because the FDE job is closer to startup reality than most corporate engineering jobs. We see that last path more often than the first two at Series A and B companies.

Do FDEs burn out?

More than pure product engineers, noticeably. The all-in job is intense. The best FDE programs rotate people out of customer-facing work every 18 to 24 months and into internal platform or tooling work for a cycle. Programs that do not rotate people tend to lose their senior FDEs to competitors by month 30.

Is this role getting automated?

The boring half of it, yes. Customer-specific customization work has shrunk at companies that invested heavily in configuration APIs and templated deployments, and that trend accelerates every quarter. The parts that require reading a VP’s tone of voice in a meeting are not getting automated this decade. Most FDE work is the second category.

When to Open the Forward Deployed Engineer Req

Three signals that your company actually needs FDEs and not some adjacent role.

Your first enterprise customer’s deployment is eating 30% of your engineering team’s time. That is the moment to stop pretending product engineers can do this work.

Your sales team keeps closing deals that include custom integration work, and customer success does not have engineers on staff. The deal economics stop making sense without FDEs.

Your product has reached the scale where design partnerships are how new features get validated. Product FDEs are the right hire for that stage, and they should be senior.

If those conditions do not apply, you probably want a solutions engineer or a customer success engineer instead. Posting an FDE req when you need one of those roles costs months.

For companies in Southern California or the broader US market looking to staff the role, our team has placed FDEs at AI startups, enterprise SaaS, and data platform companies over the last 18 months. Reach out to our team if you want a benchmark on comp, a JD review, or a short list of pre-qualified candidates. First conversation takes about 20 minutes.