Responsible for the customer's results. No authority to make them happen.

Ted Mabrey runs Palantir's commercial business. When he describes what his company expects from a Forward Deployed Engineer, one sentence stands out: FDEs should hold themselves accountable for the customer's business results and "act as if you are the CEO, but with zero authority."

The debate about FDEs usually gets framed as a question of where the role belongs. A rebranded Sales Engineer with a bigger salary? Professional services with production access? Product management on the road? Hardly anyone asks what that sentence demands of the person who chooses this role.

I coach Sales Engineers and technical leaders. That person is who this article is about.

What the job actually asks

Henley Wing Chiu at Bloomberry analyzed 1,000 FDE job postings. Working directly with customers was the most common responsibility, named in 55% of them. Building and deploying AI systems came second at 37%, integrating systems and APIs third at 32%. Nearly half the postings explicitly asked for customer-facing ability.

Chiu also lists what separates the people who succeed. Plenty of engineers can write Python and deploy models. Far fewer can adapt to a new customer's industry and tech stack every few months, convince a VP why things need to be done a certain way, and own a deployment when it breaks at 2 AM.

Nabeel Qureshi spent eight years as an FDE at Palantir. FDEs there were expected to work from the customer's offices three to four days a week. For his first engagement, he moved to Toulouse for a year and worked in the Airbus factory, alongside the manufacturing teams, four days a week. What the job required, in his description: high pain tolerance, and the social and political skills to embed deep in a foreign company and earn its trust.

Jeff Darnton, quoted by LeadDev, puts it more simply: soft skills and the ability to talk to customers are the strongest separator for an FDE.

Read those descriptions again and notice what they add up to: a way of living, inside another company's building and its crises, for months at a time.

For some people, that is exactly the point. Qureshi writes that he joined Palantir to work on hard, meaningful problems in difficult industries. People like him choose the pressure, because it fits what they care about. What carries them is rarely a skill. It is a mindset and an inner set of values that turn a high-pressure situation into meaning: finding purpose in someone else's success and feeling aligned with what the work asks of them.

The job postings list skills. Whether you thrive in the job depends on whether the pressure fits your values.

Responsible for everything, in charge of nothing

Back to Mabrey. In his description, the FDE owns every reason the customer might fail as a gap the product has to close. Much of the work is non-technical: organizational alignment, user adoption, rethinking how a business process runs. Most technology companies push those problems onto the customer. The FDE takes them on personally.

KDnuggets names the cost directly. Constant context switching, months as the company's face at a client, and accountability for adoption the FDE doesn't control make burnout a structural risk of the role, not an individual weakness.

Research on sales roles has a name for this: role conflict. Being held to expectations that pull in different directions. A study by Boles and colleagues in the Journal of Personal Selling & Sales Management found that role conflict and work-family conflict were both significantly related to emotional exhaustion, and exhaustion in turn to people's intention to leave. That study is about salespeople, not FDEs. But look at the FDE's position: loyal to the customer's success, to their own company's product, and to a life somewhere else, all at the same time.

There is a sharper version of the conflict inside the work itself. Mabrey writes that the FDE welcomes scope creep, because the customer's mission demands it. Others, among them Tim Brömme on the PreSales Unleashed podcast, stress the opposite skill: saying no to customizations that help one customer and nobody else. Both are right in their context. Which one applies today, in this meeting, with this sponsor, is decided by one person in the room. Without the authority to make it stick.

Accountability without authority is a classic recipe for role conflict. For an FDE, it is written into the job.

FDE or SE: the debate, and the person in the middle

Start with the organizational question.

Bloomberry's data separates the roles clearly on paper. A Solutions Engineer's job ends when the customer says yes, this will work for us. An FDE's job starts when the customer says yes, now make it work. None of the 1,000 postings carried a quota. 45% placed the FDE in a dedicated team, 38% in engineering, only 14% in sales or go-to-market. 68% required travel. Chiu also found that roughly three in ten postings are Sales Engineer roles with a new title, quota and commission included.

a16z's Joe Schmidt explains why companies pay for all this. Enterprises buying AI are like a grandparent getting an iPhone: they want to use it, and they need someone to set it up. Vendors accept lower margins on that hands-on work because it makes them hard to replace. Trading margin for moat, in his words.

From the presales side, Tim Brömme argues in episode 281 of PreSales Unleashed that the customer half of the FDE job is the work good SEs already do: discovery, stakeholder management, managing expectations. The engineering is the entry ticket.

Each answer puts the role in a different box on the org chart. What lands on the person stays the same. Whichever box you choose, one human being carries the engineer's standard (the code has to run in production), the advisor's duty (the customer has to trust you) and the product team's need (what you build should help more than this one customer).

And the account load makes it personal. An SE spreads attention across many deals. An FDE goes deep into one or two customers. When that customer has a crisis, it becomes your whole job. When you sit inside their organization for weeks, their pressure becomes your pressure, and your own team is somewhere else.

Where it goes wrong

When nothing moves

Qureshi describes how data access was often blocked less by technology than by politics: teams whose standing in the company rested on being the gatekeepers of a data source. Some pilots of eight to twelve weeks went entirely into getting access, with the final week spent scrambling for something to demo.

Now look at it from the inside. You are on site. Every day you walk into the same building and nothing moves. The customer sees you, so the pressure lands on you. And if you are the kind of engineer who got good by solving things, you do what always worked: you work harder on the part you can control. Late nights and a rushed demo. And a quiet sense of having failed, although the cause sat in someone else's inbox.

When you're boxed in

Mabrey adds a different kind of failure. Former Palantir FDEs who joined companies copying the model tell him it feels like being in jail: the business strategy decides what they are allowed to take on, so they carry the customer's expectations without the room to meet them.

Here the pull goes the other way. Delivering exactly what was asked feels safe. It is measurable, and nobody can blame you. Pushing back on a customer who pays for your time means risking that you look unhelpful or difficult. So the yes often comes out before you have even decided.

When it never stops

Piotr Kraus, a former Palantir FDE who now advises companies on the model, told LeadDev that constant travel makes the role susceptible to burnout much earlier than others. Recruiter Adam Moore adds, in the same piece, that it is definitely not for everybody.

I know a version of this from my own career. For more than two years I worked for IBM Strategic Outsourcing in Basel, Switzerland, 650 kilometers from my hometown. I traveled back almost every weekend and was still away more than I was at home. We were migrating data centers, and I sat between at least two sides: the customer's team and IBM's delivery team. Whether I said yes or no to either side was rarely an architecture decision. It was a political balancing act.

I learned a lot in those two years. I also lost something, being constantly on the road and abroad. Life somewhere else doesn't wait for you.

In each of these cases, the person on site absorbs a problem that was never theirs alone.

What the skill lists leave out

On the skills, the sources largely agree. Discovery that reaches the real problem. Success criteria agreed before anything gets built. Knowing who owns the data and who can block the project. Explaining a setback to an executive in their terms. Declining the customization that helps nobody else. These can be learned, and good training for them matters.

Underneath each of them sits something a skills list doesn't reach.

You can learn the words for a good no. Whether they come out in the room depends on what you believe a no will cost you. If somewhere inside you the rule is "my value is what I deliver," every no feels like a small risk to your worth. The framework is in your head. The yes comes out anyway.

The same goes for executive communication. You can know exactly how to report a setback to a VP. If part of you hears that setback as proof that you are not good enough, you will over-explain, add one more technical detail, and lose the room you were trying to hold.

Training installs the method. Whether you can use it under pressure is a question about who you are in that moment.

That is the work coaching does. We look at the patterns that took you this far: the drive to solve everything yourself, the need to be the one with the answer. Those strengths made you a great engineer. In a role this exposed, with this much responsibility and this little authority, the same strengths can become the ceiling.

What it takes to stay

The pay reflects how much is asked. Bloomberry puts the median FDE base salary at around $174,000 across postings that disclosed it, with 70% offering equity on top, and the AI labs pay well above that.

The money helps. You still carry the weight. Months on the road, one customer's crisis as your whole job, responsibility without authority, and failure modes where you carry problems that were never yours alone. Over months, that combination wears people down.

The people who last in roles like this tend to have four things in place. They know why they chose the role, and the work fits their values. They know where their limits are and say so early. They have a space outside the customer where they can think out loud, about the deployment and about themselves. And they have learned to separate their worth from the outcome of a single engagement.

If you lead a team that is standing up FDEs, plan for the person as carefully as you plan for the skills. Who do they debrief with after a hard week on site? How often do they come home to their own team? Who backs them when they push back on a customer you would rather keep happy?

If you are the one considering the role: what would you need in place to act like the CEO, with zero authority, for two years, without losing yourself in it?

Sources