| Company | Base salary | Experience | Travel or location |
|---|---|---|---|
| Palantir FDSE, NYC | $135k to $200k | 1+ year | Up to 25% |
| OpenAI FDE, SF | $162k to $280k + equity | 5+ years | Up to 50% |
| OpenAI FD Software Eng, SF | $185k to $325k + equity | 7+ years | Up to 50% |
| Sierra FD Infrastructure | $230k to $390k + equity | 4+ years | In person, SF |
| Ramp FD AI, NYC | $189k to $330k + equity | Not stated | Up to 75% |
| Baseten | $165k to $330k + equity | 2+ years | Not stated |
| Scale AI | $179k to $224k | 2+ years | Hybrid, 3 days, SF or NYC |
| Redpanda | Not published | Not stated | Posted “US Remote”, asks 30 to 50% |
| Gradial Seattle | Not published | 1+ year | Under 10% |
Read from the companies’ own job boards on 6 August 2026. Bands are base salary for US roles and exclude equity unless noted.
Two years ago almost nobody outside Palantir used the phrase. Now OpenAI runs a forward deployed engineering department with its own hiring manager and its own recruiter, and the Wikipedia article explaining the term, created on 5 May 2026, was read 23,524 times in July. The DevOps article managed 17,760 in the same month.
So it is worth knowing what the job actually is before deciding whether you want it. The table above is the honest version, taken from the companies’ own postings rather than from a recruiter’s summary.
Three things stand out immediately. The money is real. The experience bar is all over the place, from one year to seven. And the travel column is the one that decides whether you would last.
You work inside somebody else’s company, on their systems, solving their problem, using your employer’s product as the raw material.
Redpanda’s posting says it about as plainly as anyone: embedded on-site, working inside the customer’s environment, writing the code that ships. The same posting adds a sentence that tells you what the market is worried about, which is that this is not a solutions engineering role.
The tools are ordinary. Python, TypeScript, cloud infrastructure, the usual. Nothing in the stack explains the salaries.
What you’re paid for is the willingness to sit inside a stranger’s organisation, alone, and still ship.
That means reading a codebase nobody documented, in a company whose politics you don’t know, while the person paying for the project watches you do it. Then explaining, to somebody senior and impatient, why the thing they asked for on Monday turns out to be the wrong thing.
The bands in the table run from $135,000 to $390,000, and several list equity on top. Sierra’s infrastructure role leads at $230,000 to $390,000, and the company says openly that it is primarily an in-person company based in San Francisco.
That spread is not a market disagreement about what the job is worth. It is three different jobs sharing a title.
Palantir asks for one year of experience and pays from $135,000. OpenAI’s senior version asks for seven years of full stack work and explicitly values having been a founder or an early engineer who built a product from scratch. Those are not the same role with different numbers on it.
Now the column people skip. Travel is the price.
Seventy-five percent means three weeks in four somewhere else. Read that number before you read the salary, because it’s the part that decides whether the job is compatible with your life, and it’s the part that no recruiter leads with.
The Redpanda line deserves its own mention. A role advertised as remote that still wants half your weeks on a customer’s site is not a lie, but it is a trap for anyone who skims.
The title is hot, so it’s being stuck on roles that are not this. Sales engineering, solutions consulting, and customer success all now appear as “forward deployed” somewhere. Three tests separate them.
Does your code reach production? A real posting talks about building and deploying inside the customer’s environment. A relabelled one talks about demos, proofs of concept, and supporting the account team. Demos are a fine job. They’re a different job.
Where does the salary band sit? Compare it with the ordinary engineering ladder at the same company. If the FDE band is below the software engineer band, you’re looking at a customer-facing role wearing an engineering title.
Who carries the commercial side? Ramp’s posting is unusually specific about this. The engineer co-leads each customer engagement with a solutions strategist, which tells you the account has an owner and that it isn’t you. Most postings say nothing at all, so ask in the first call. If nobody else owns the relationship, you have found the sales job.
Take it if you like finishing things, and if the part of product work that bores you is shipping a feature into a void where nobody tells you what happened. Here you watch someone use it, badly, in front of you, and you fix it that afternoon.
Take it if ambiguity wakes you up. No backlog groomed by somebody else. The problem arrives as a complaint from a person who doesn’t know what is technically possible, and turning that into a spec is most of the job.
Take it if you want to see how large organisations really work. That knowledge compounds, and there is nowhere else to get it.
Now the other side. Deep focus is off the table, because the work is interrupt-driven by design.
So is owning a system for years. You build, you hand over, you leave. Somebody else maintains it, often worse than you would, and you have to be at peace with that.
And if travel is a hard no, believe the number in the table rather than the word remote at the top of the page.
The experience bar is lower than the salaries suggest, which is the most useful thing in the table. Palantir wants one year. Gradial wants one year. Baseten and Scale AI want two.
What they filter on instead is evidence that you’ve worked with people outside engineering and survived it. So the things worth putting in front of them are the unglamorous ones: the integration you ran with a customer’s team, the migration where you negotiated scope with somebody’s operations lead, the project where the requirements changed because you asked a question in week one.
Consulting experience helps here more than another framework does. So does having been the technical half of a very small company, which is why OpenAI names founders explicitly.
One practical note if you’re weighing offers. Ask what happens after the deployment ends. Some companies rotate you onto the next customer. Some fold you into the product team that inherits your work. That answer tells you whether you’re joining an engineering organisation or a delivery arm, and the two feel very different by year two.
The AI labs made the role visible and it has spread from there. Look at the table again and notice how few of those companies are labs. A streaming data platform, a payments company, a model hosting startup, a company nobody outside Seattle has heard of. That is what a title looks like on its way into the mainstream.
Whether the term itself survives is a separate question. Enough people already read it as a renamed sales engineer that the meaning is under strain, and titles stretched that far usually snap. The work isn’t going anywhere though. The thing driving it is real: software that needs shaping to each customer sells better when somebody technical does the shaping on site.
If you’re the kind of engineer this describes and you’d rather do it on remote-friendly terms, tell us what you build. We place senior engineers with startups that need someone to own an outcome rather than close a ticket.
Privacy is important to us, so you have the option of disabling certain types of storage that may not be necessary for the basic functioning of the website. Blocking categories may impact your experience on the website. More information
