FDE vs Solutions Architect vs Sales Engineer, by the numbers

Two numbers — how much you code and how much you sell — split this job family in half. Here is where each title lands, and how far apart two postings sharing a title can be.

August 17, 2026

Five job titles, one job family, and no agreement anywhere about which is which. A “Solutions Architect” at one company writes production code four days a week; at another, the same title is a pre-sales role that opens an IDE twice a quarter. The advice usually offered — that FDEs code more, SEs sell more — is directionally right and practically useless, because it can’t tell you which one the posting in front of you is.

This board classifies every posting it publishes, scoring each one on seven dimensions from the job description itself rather than from its title. That produces something the usual title debate can’t: a measurement. Below is what 1,150 live postings say about where each of these five roles actually sits — and about which titles you can trust.

The two numbers that separate the roles

Of the seven scored dimensions, two do nearly all the work of telling these roles apart: coding intensity (how much of the job is writing software) and sales involvement (how much of it is winning the deal). Plot the median of each and the family splits cleanly in two.

The gap is not subtle. Forward Deployed and Applied AI roles cluster at coding 75–80 with sales involvement of 15. Solutions Architect and Sales Engineer sit at coding 35 with sales at 65 and 75. Implementation / Delivery lands in its own corner: low on selling, but also well below the FDE cluster on code, because much of the work is configuration, integration, and project sequencing rather than building.

Two things follow. First, FDE and Applied AI Engineer are close to the same job on this measurement — the difference between them is subject matter, not shape. Second, Solutions Architect is much nearer to Sales Engineer than to FDE, which is the opposite of what the word “architect” suggests to most engineers reading the posting.

All seven dimensions

Median score per dimension, 0–100. “Customer” is customer-facing, “Complexity” implementation complexity, “AI use” AI/LLM usage. Postings are counted under their primary label; 576 of the 1,150 carry a second label as well.

Note what doesn’t separate them. Customer-facing scores sit between 75 and 90 for all five — every role in this family is client-facing, so “customer-facing” is a property of the family, not a way to choose within it. If a recruiter distinguishes two of these roles by telling you one is more customer-facing, they have told you nothing.

The title lies more often than you would guess

Because the classifier reads the description rather than the title, it can be asked how often the two disagree. For the postings it labelled:

• Forward Deployed Engineer — 8% of postings don’t say anything like “forward deployed” or “FDE” in the title.
• Solutions Architect — 18%. Sales / Solutions Engineer — 17%.
• Applied AI Engineer — 68%. Most of these are titled “Software Engineer, Agent”, “AI Engineer”, or some product-specific variant.
• Implementation / Delivery — 76%. This work hides under “Technical Support Engineer”, “Value Engineer”, “Designated Support Engineer”, “Delivery Solutions Architect”, and a dozen other constructions.

So title search works acceptably for FDE and badly for everything else. If you are searching LinkedIn for “Applied AI Engineer”, you are missing roughly two thirds of the roles that are, in substance, exactly that job.

Same title, opposite jobs

The more useful question is not what the median posting looks like but how wide the spread is inside one title. Take every posting with “Solutions Architect” in the title, whatever the classifier concluded:

• 40% score 70 or above on sales involvement — pre-sales roles in all but name.
• 13% score 30 or below — genuine post-sales delivery work.
• Coding intensity runs from a 10th percentile of 25 to a 90th of 65. A quarter of them (27%) sit at 30 or under.

Those are not variations on one job. They are two different careers sharing a phrase. For comparison, postings with “Forward Deployed” in the title are far more consistent: 86% score 70 or above on coding intensity, and only 3% score 30 or below. The FDE title, whatever else you think of it, means something specific.

Travel is the real dividing line

The dimension candidates underweight most, and the one that changes daily life most, is travel. Share of postings in each band:

• Forward Deployed Engineer: 29% high travel (>50%), 32% medium, 37% low or none. The widest spread of any role here — “FDE” tells you nothing about whether you’ll be on a plane.
• Solutions / Sales Engineer: 14% high, 62% medium.
• Solutions Architect: 5% high, but 80% medium — the most uniformly travel-bearing role of the five, and the one least likely to be either extreme.
• Implementation / Delivery and Applied AI Engineer: 5% and 3% high, with 42% and 41% low or none. If you want this kind of work without the airports, these two are where to look.

Which of these roles actually touch AI

Every company in this market claims to be an AI company. The descriptions are less enthusiastic. Share of postings scoring 70 or above on AI/LLM usage:

• Applied AI Engineer: 96% (median 85).
• Forward Deployed Engineer: 49% (median 65).
• Sales Engineer 29%, Implementation 26%, Solutions Architect 18% — all with a median of 40.

The tech stacks say the same thing. FDE and Solutions Architect postings are dominated by the data platform and cloud — Python, AWS, Azure, GCP, Spark, Databricks — while Applied AI postings are the only group where LLMs, agent frameworks, and RAG pipelines appear near the top of the list. If you want to work on frontier AI rather than alongside a company that sells it, Applied AI Engineer is the label that reliably delivers it, and FDE delivers it about half the time.

What we are not claiming

Three limits worth stating plainly, because they bound how far these numbers travel.

• Employers are concentrated. The five biggest posters account for a large share of the sample — Databricks alone posts 111 of the 275 Solutions Architect roles. So the medians could in principle be describing a handful of companies rather than a market. Two checks say they aren’t: giving each employer one vote (the median of per-employer medians, employers with three or more postings) moves coding intensity and sales involvement by at most five points, and dropping Databricks entirely moves them by at most five points as well. The shape of the chart survives both.
• No salary figures here. Only a few hundred postings disclose a range, and the disclosures are worse than concentrated — 43 of the 57 FDE postings with a US range carry the same range from one employer. A median computed on that would be one company’s pay band wearing a market’s clothes. Pay deserves its own piece, with the disclosure problem handled honestly.
• These are model scores, not ground truth. An LLM classifier reads each description and scores it; low-confidence results are routed to human review before publication, but the scores describe what the posting says, not what the job turns out to be. Employers routinely undersell travel and oversell coding.

How to read a posting after this

The practical upshot is that the title should be the last thing you weigh, not the first. Four questions get you the rest of the way:

• Does the description name a quota, a territory, a pipeline, or an account executive? That is the single most reliable pre-sales tell, and it moves the role from the left of the chart to the right regardless of what the title says.
• Does it name a language or a stack in the requirements, or only “technical aptitude”? Coding intensity above 70 nearly always comes with specific technology named in the posting.
• Does it say post-sales, deployment, onboarding, or time-to-value? That is implementation work, whatever the title — and it is its own career track, not a junior version of the others.
• What does it say about travel — and does the number appear at all? A posting that omits it is not promising you zero.

If you’d rather not do that parsing by hand, every posting on this board is already scored on these seven dimensions, and the filters run on the scores rather than on the titles. Start from Forward Deployed Engineer, Solutions Architect, or Solutions / Sales Engineer and narrow from there. If you’re preparing for a loop rather than choosing a role, the FDE interview guide covers what each stage is scoring, and the role guide is the plain-English version of this comparison.

Method

Every job published on DeployedJobs as of 17 August 2026 was read through the public API: 1,150 of 1,169 postings were retrieved successfully, of which 1,129 carry a primary role label. Each posting’s seven dimensions are scored 0–100 by an LLM classifier at ingestion time, from the description text alone; results below a confidence threshold go to a human review queue rather than straight to publication. Figures quoted are medians unless stated otherwise, and postings are grouped by their primary label — 576 postings carry a secondary label too, most commonly Solutions Architect alongside Sales Engineer (86 postings) and Implementation alongside Solutions Architect (82).

Browse Forward Deployed Engineer jobs