Forward Deployed Engineer vs Solutions Architect: Which One Do I Actually Need?
Hire a solutions architect when your customers have engineers who will do the building, and a forward deployed engineer when they do not. The architect designs the system and hands the design over. The forward deployed engineer goes inside the customer's environment, writes the code that runs there, and is still accountable for it a quarter later. Neither is the senior version of the other. They are different purchases, and the question that separates them is not how good the person is. It is who is going to build the thing.
The market has already answered this, and you can read the answer
I get asked this question in the abstract and I think that is why it stays confusing. So I went and looked at what the companies who employ both roles are actually doing with their money this week. Three of them, three different answers, all published openly.
Palantir invented this job. On the morning I wrote this it had 309 open positions. Seventy six of them carry forward deployed in the title, 21 of those in Washington, 16 in New York and 12 in London. The number carrying architect in the title is zero. Not few. Zero.
Anthropic, which is two years into building the same motion, had 569 open positions and a Forward Deployed Engineer role covering New York City, San Francisco and Seattle. Solutions architect roles: also none.
Then Amazon Web Services, which is the opposite case. A search of its careers site for United States roles returns 632 matches for solutions architect and 17 for forward deployed. Both organisations exist, side by side, hiring at the same time, and the ratio is about thirty seven to one in the architect's favour.
Those three facts look contradictory until you ask what each company sells.
The real dividing line is who does the building
Amazon Web Services sells a catalogue. Hundreds of thousands of customers implement it themselves, with their own engineers, and what those engineers need is someone who knows the catalogue better than they do and will tell them which four services to combine. That is a wonderful, high leverage job, and one person can do it for dozens of accounts at once. The Startup Solutions Architect posting says the quiet part out loud: founders need a trusted technical advisor in their corner. Advisor. The founder's own team writes the code.
Palantir sells something no customer has ever implemented before, into organisations whose engineering capacity is committed elsewhere and whose data is a genuine mess. Advice is worthless there. Somebody has to go and do it. The Palantir posting in New York asks the engineer to own the end to end execution and implementation of high stakes projects, and to expect to travel up to twenty five percent of the time to be physically where the problem is. Anthropic's version asks its engineers to work within customer systems to build production applications and to provide white glove deployment support inside enterprise environments, with the same twenty five percent travel estimate.
So the test is not about your product, your stage or your funding. It is one question about your customers. When the design is finished, who writes the code? If the answer is a team on their payroll, you need an architect and a forward deployed engineer would be an expensive way to do the same job once instead of forty times. If the answer is nobody, an architecture document is a beautifully argued description of work that will not happen.
Note that Amazon Web Services opened a forward deployed engineering organisation anyway. Seventeen roles against 632 is not a pivot. It is a company with a catalogue admitting that in some accounts the catalogue is not enough and someone has to go in.
What each one costs in the United States
The comparison is usually made across companies, which tells you very little because it mixes in how rich the employer is. Amazon Web Services publishes bands for both roles, so you can hold the employer constant and read the difference cleanly.
In New York, the Startup Solutions Architect band is $144,500 to $195,400. The Forward Deployed Engineer band on the government team is $158,100 to $213,800. That is about nine percent more at the bottom of the range and about nine percent more at the top, for the same employer, the same city and roughly the same level. The senior forward deployed engineer posting sits at $184,900 to $250,200 in New York and $193,300 to $261,500 in San Francisco and Mountain View. Amazon adds sign on payments and restricted stock on top of all of those.
At the frontier labs the premium is much larger. Anthropic's Forward Deployed Engineer role lists $280,000 to $320,000 for New York City, San Francisco and Seattle. It asks for four or more years in a technical customer facing role. Palantir, interestingly, asks for one or more years of post college experience for its New York Forward Deployed Software Engineer, which tells you it is buying raw judgment and training the rest.
Read those numbers as a chief financial officer and the nine percent is not the interesting part. The interesting part is coverage. One architect at $195,400 can influence forty accounts. One forward deployed engineer at $213,800 covers one, sometimes two. On a cost per account basis the forward deployed engineer is not nine percent more expensive. They are an order of magnitude more expensive, and worth it precisely when the alternative is an account that never goes live.
These are American bands. The same roles in London price lower and the argument does not change at all.
Where the solutions architect wins
Plainly, and more often than the current excitement about forward deployment lets people admit.
The architect scales and the forward deployed engineer does not. That is the whole case and it is a strong one. If you are solving the same problem for the fortieth time, the correct response is a reference architecture, a pattern library and one very good person who keeps forty teams from making the same mistake. Sending an engineer into each of those forty accounts to hand build the answer is not thoroughness, it is a refusal to productise, and the margin will tell you so within a year.
The architect is also the more predictable hire. It is a settled role with an understood ladder, a certification path and two decades of people who have done it. Forward deployed engineer is roughly two years into a boom, which means the title is doing very different jobs at different companies and you will spend real effort working out what a given candidate actually did. Job listings for it grew around 800 percent between January and September 2025 and more than 1,000 percent year on year into early 2026. Titles do not stay precise under that kind of pressure.
And there is a version of the forward deployed engineer that fails exactly where the architect would have won. They build you a perfect thing for one customer that cannot be reused for the next one, because nobody was holding the view across accounts. That is the architect's actual job, and losing it is expensive in a way that shows up two years later in your cost of delivery.
Where the forward deployed engineer wins
When the design is not the bottleneck. This is the failure I get called into, and from the inside it never looks like an engineering problem. It looks like a signed customer who has gone quiet. There is an architecture document. Everybody agreed with it. Nothing is running.
What happened is almost always the same thing. The plan assumed an implementation team on the customer's side, and that team either does not exist, or exists and is busy with the thing their chief executive actually cares about. The design was never wrong. It was addressed to somebody who was not going to read it.
The companies with the best products in the world reached this conclusion with their own balance sheets. In May 2026 OpenAI launched a four billion dollar deployment subsidiary. In July 2026 Anthropic launched Ode with Anthropic, built on its acquisition of Fractional AI and backed by Blackstone and Hellman and Friedman. Google Cloud is hiring 59 forward deployed engineers in 2026. None of those companies has a design problem. They discovered that a superb product does not install itself into an insurance company's claims process, and no quantity of architecture fixes that, because the gap opens after the document is delivered.
The mistake I see most
Buying an architecture when what you needed was a builder. It is the easier purchase to approve. It costs less, it arrives faster, it produces a document you can show the board, and everyone in the room agrees with it, which feels like progress and is the single most misleading signal in this whole decision. Agreement is cheap. Nobody disagrees with a diagram.
The reverse happens too and it is less costly but more embarrassing. A forward deployed engineer is dropped into an account that had four other accounts exactly like it, builds a bespoke solution beautifully, and now you own five bespoke solutions and no product.
There is a third mistake, quieter than both. Title inflation. At a thirty person company, solutions architect frequently means a forward deployed engineer with a title that was easier to recruit against, and occasionally the reverse. When I am reading a job description, mine or somebody else's, I ignore the title entirely and look for one sentence: does this person get write access to a system they do not own? If yes, it is forward deployment whatever it says at the top. If no, it is architecture, and it should be priced and staffed as architecture.
Twenty years running finance and technology inside high growth companies, through three exits, taught me to read this as a cash flow question rather than an org chart question. An architect converts confusion into a plan. A forward deployed engineer converts a plan into revenue. Buy whichever conversion is currently blocked, and be honest with yourself about which one that is, because most companies I meet have a drawer full of plans.
One question that settles it in a meeting
Take your last three stalled accounts and ask what would have to be true for each of them to be live next month. If the answer is a decision, you need an architect. If the answer is somebody sitting down and doing four weeks of unglamorous integration work against a schema nobody documented, you need a forward deployed engineer, and you needed them a quarter ago.
Almost nobody answers decision three times. That is what the hiring data at Palantir and Anthropic is telling you, and it is why Amazon Web Services started an FDE organisation despite having 632 architects.
Where I come in
At Beyond Elevation I take the second chair, not the first. Hayat Amin, forward deployed into the company, with write access to the systems and my name against something that has to run in production, which is the arrangement I would want if I were the one paying. It is deliberately not an architecture practice. The fractional version is one or two days a week from $5,800 a month, and a scoped project with an end date runs eight to fourteen weeks from $30,000, because a builder with no end date slowly becomes an advisor.
For founders in New York and across the United States sitting on this exact decision, the useful half hour is not a debate about two job descriptions. It is going through your stalled accounts one at a time and marking each one decision or build, because that list picks the hire faster than any comparison, this one included. Book that at meethayat.com/services/fde. If the blockage is money rather than deployment, the CFO seat is at meethayat.com/cfo. The closest neighbour to this question is forward deployed engineer vs solutions engineer, which turns on the contract rather than the build, and if you suspect the whole category is a rebrand I answered that in is a forward deployed engineer just a consultant. Otherwise start at meethayat.com.
Questions I get asked about this
Is architect better than engineer?
No, and the assumption that architect is the senior version is what makes companies buy the wrong one. They are parallel roles with different products. The architect's product is a decision other people can act on. The forward deployed engineer's product is a running system. At Amazon Web Services the two sit in separate organisations and are hired against separate ladders, which is the clearest evidence available that neither is a promotion from the other.
Cloud engineer vs solutions architect: where does the forward deployed engineer sit?
The cloud engineer runs your own infrastructure and faces inward. The architect faces a customer and designs. The forward deployed engineer faces a customer and builds, inside that customer's systems. The useful axis is not technical depth, since all three have it. It is whose production environment the person is trusted to change. A cloud engineer changes yours, an architect changes nobody's, and a forward deployed engineer changes the customer's.
Can a solutions architect become a forward deployed engineer?
Yes, and it is more about appetite than skill. The architecture instinct transfers whole. What changes is tolerance for other people's mess, because the forward deployed engineer inherits the real data, the real permissions and the legacy system that was not in the diagram. The architects who make the move well are the ones who were already frustrated watching their designs implemented badly by somebody else.
Do I need a forward deployed engineer or a solutions architect for my startup?
Count how many of your customers have engineers who will do the work. If most do, buy the architect, because one architect can guide many implementations and a forward deployed engineer can only be in one place at a time. If most do not, buy the forward deployed engineer, because a plan handed to a team that does not exist is not a plan. Below Series A almost nobody needs both, and the companies that hire both early usually needed one senior person who could do either.
Where do these numbers come from?
Every figure above was read on 1 September 2026 from the live sources themselves rather than from a summary. The posting counts and salary bands come from Palantir's careers feed, Anthropic's careers board and the Amazon Web Services careers site. The growth figures for forward deployed engineer hiring, and the notes on OpenAI's deployment subsidiary, Ode with Anthropic and Google Cloud's 2026 hiring, come from reporting by Pragmatic Engineer, Perspective AI and Paraform. Job boards move. If you are reading this months later, the bands will have shifted and the ratios are the part worth keeping.