FDE vs Software
Engineer
Both write production code. Both get paged. The difference is not seniority, and it is not "one talks to customers" — it is who owns the outcome, and over what time horizon. That single change rewrites the day, the comp curve, and what good looks like.
The one-sentence delta
A product software engineer builds a feature for many users, over quarters, behind a roadmap, and is done when it ships correctly. A forward deployed engineer builds a solution for one customer, over weeks, against a problem nobody wrote down, and is done when that customer’s workflow actually changed.
Everything below is a consequence of that sentence.
Notice what the delta is not. It is not code volume — FDEs write real production code; roughly 40% of the job at a frontier lab is building.10 It is not seniority — hiring guides describe the FDE target profile as staff-engineer material.11 And it is not "FDE is customer-facing" on its own; plenty of product engineers talk to users. The delta is ownership scope: an FDE cannot hand the outcome to a PM, an SRE, a solutions architect or a support org, because at the customer site those people are the same person.
Side by side
| Dimension | Product software engineer | Forward deployed engineer |
|---|---|---|
| Optimises for | Scale, correctness, maintainability | Customer value shipped, fast |
| Users | Many, anonymous, aggregated in dashboards | Few, named, in the room with you |
| Time horizon | Quarters | Days to weeks |
| Spec comes from | PM, roadmap, design doc | You, after watching people work |
| Environment | Your infra, your CI, your observability | Their cloud, their firewall, their auth, their 2011 database |
| Definition of done | Merged, tested, rolled out | Adopted, measured, handed off |
| Code lifetime | Years; abstraction pays | Weeks to years; a throwaway prototype is a legitimate deliverable |
| Failure mode | Shipped the feature, nobody used it — product’s problem | Shipped the feature, nobody used it — your problem |
| Hard part | The system | The organisation around the system |
| Measured by | Velocity, quality, roadmap delivery | Adoption, renewal, whether the number moved |
“Shipped but unused” is a product failure for a SWE and a personal failure for an FDE. If that asymmetry sounds unfair, you want product engineering. If it sounds like finally being allowed to fix the whole problem, you want FDE work.
A week in each
The clearest way to feel the difference is to put two calendars next to each other.
Product SWE — feature week
- Mon Design review for the new export pipeline; comment on two RFCs.
- Tue Implement the happy path; three-hour uninterrupted block.
- Wed Edge cases, tests, a flaky CI job.
- Thu Code review round trips; feature-flag rollout to 5%.
- Fri Watch dashboards; write the launch note; pick up the next ticket.
FDE — week three of a deployment
- Mon Shadow three analysts. Discover the “report generation” ask is really a five-systems data problem.
- Tue Get read access to two of the five. Start the security review for the other three now, because it takes six weeks.
- Wed Hardcoded prototype over the two you can reach. Ugly. Works on one real record.
- Thu Demo that one record to the champion. Their reaction reshapes scope. Capture five new “can it also…”.
- Fri Third system returns 403 from inside their network. Egress rules. Two hours with their network team, then write the eval that will define “working”.
Two things jump out. First, the FDE week has no three-hour block — and that is the single most common reason experienced engineers bounce off the role. Second, the FDE week contains at least three tasks that would be somebody else’s job in a product org: discovery, procurement/security, and defining the acceptance criteria.
The compensating pleasure is closure. A product engineer often never learns whether the feature mattered. An FDE watches an analyst stop doing four hours of manual work, on Thursday, in person. Nabeel Qureshi’s account of forward-deployed work at Palantir — four days a week on a customer site for a year — is the canonical description of both halves of that trade.14
Compensation
At the same company and level, FDE comp is usually equal or higher — but the distribution is much wider and the equity share is larger.
| Data point | Figure | Confidence |
|---|---|---|
| Median posted base, strict-definition FDE roles3 | $185K | Job-posting census |
| Palantir FDSE total comp16 | $171K–$295K, median ~$211K | Self-reported aggregate |
| Frontier-lab FDE, mid-to-senior18 | Reported $350K–$550K, rising past $1M at staff level | Survey, self-reported |
| Equity share of total comp at frontier labs18 | Reported 55–70% | Survey, self-reported |
Three structural reasons FDE comp runs hot right now, none of which are permanent laws:
Scarcity of the combination
Strong engineer and willing to sit with a customer is a rarer intersection than either trait alone. Demand for the combination grew 729% year over year on one index.1
Direct line to revenue
An FDE is attached to a named account with a contract value. When your work visibly protects a seven-figure renewal, the comp conversation is different from “shipped 14 tickets”.
It is a harder job to fill than to describe
Companies discovered that hiring a junior into the seat produces a stalled deployment, so the bar moved up to staff-level judgement11 — and comp followed the bar.
The wide distribution cuts both ways. A pre-sales-shaped “FDE” role at a mid-market vendor can pay below a normal senior SWE offer and carry a variable component. Roughly 30% of postings with the FDE title are that shape.1 Read the comp structure before the title. Full salary breakdown →
What transfers, what does not
Transfers straight across
- Debugging instinct — the single most valuable import.
- API design, data modelling, SQL.
- Containers, one cloud, CI, secrets hygiene.
- Knowing what production actually requires.
- Reading unfamiliar code fast.
Has to be built new
- Scoping under ambiguity — deciding what not to build.
- Evals as acceptance criteria — turning “it should be accurate” into a gate.
- Deploying where you do not own the cloud: VPC, on-prem, air-gapped.
- Narrating your thinking to a sceptical non-engineer in real time.
- Saying no in a way that increases trust.
One habit actively hurts you: the instinct to build the general, correct abstraction first. In product engineering that instinct is a virtue. In week one of a deployment it burns the only currency you have — time before the customer’s attention moves on. FDEs learn to hold it until the problem is understood, then spend it on the two or three pieces that will outlive the engagement.
The four reflexes a SWE is usually missing
These are the parts you cannot read your way into. Each one is an interactive drill on this site — no signup.
Should you switch?
Two honest checklists. Read them as filters, not as encouragement.
Switch to FDE if…
- You are more frustrated by a feature nobody uses than by a codebase you dislike.
- You are energised, not drained, by unblocking a stuck situation.
- You want to see the whole arc — problem to production to a person using it.
- You are comfortable being the most technical person in a room of non-engineers.
- You want the fastest available route to commercial context: contracts, ROI, renewals.
Stay a SWE if…
- Deep, uninterrupted focus is how you get satisfaction from work.
- You want to go deep on one system for years.
- You are optimising for research or hard-systems credibility.
- Travel or unpredictable weeks are a real constraint in your life.
- You dislike being accountable for outcomes you do not fully control.
The last bullet on the right deserves emphasis, because it is the honest cost of the role. An FDE is accountable for adoption, and adoption depends on politics, budget cycles and a middle manager whose team the tool makes smaller. You can do everything right and lose. People who need control proportional to accountability find that genuinely painful.
Making the move in either direction
Evidence beats titles
Hiring loops want proof you have shipped something for a real user under real constraints — not a polished clone. Do one small end-to-end engagement: find someone with a genuine workflow problem, scope it, build the smallest slice, deploy it where they actually work, and measure whether it changed anything. Write it up honestly, including what you got wrong. That single artefact outperforms a year of side projects in an FDE loop.
Then close the two skill gaps that show up in interviews: scoping out loud and evals. Both are drillable.
Easier than the reverse, with one caveat
Product teams value the customer context an FDE brings, and the exit is common. The caveat is depth: after two years of integration glue, a systems-heavy loop can catch you out. Counter it by keeping one genuinely hard technical thing per engagement — the eval harness, the retrieval quality work, the deployment automation — and being able to talk about it at depth.
If you are actively preparing, the interview shapes differ more than the job titles suggest: about half of a typical FDE loop is scenario work rather than algorithms. The FDE interview prep hub breaks down every round, and How to become an FDE is the full route in.
Quick answers
What is the difference between a forward deployed engineer and a software engineer?
A software engineer builds features for many users over quarters and is done when the code ships correctly. A forward deployed engineer builds a solution for one specific customer over weeks, inside that customer’s environment, and is done when the customer’s workflow actually changed. FDEs own discovery, integration, deployment and adoption — work that is split across PMs, SREs and support teams in a product org.
Do forward deployed engineers earn more than software engineers?
Usually equal or more at the same company and level, with a much wider distribution. A 2026 posting census put the median posted base for strict-definition FDE roles at $185K, Palantir FDSE total compensation is reported at $171K–$295K with a median near $211K, and frontier-lab FDE roles are reported far higher with equity making up most of the package. Pre-sales-shaped roles carrying the FDE title can pay less than a senior SWE offer.
Is forward deployed engineering a real engineering job?
For roughly 60% of postings, yes — those are builder roles that own production code. About 30% are pre-sales-shaped and 10% are internal platform roles. The fastest screen is whether the job description mentions quota or variable compensation, and whether you own what runs in production.
Can a software engineer become a forward deployed engineer?
Yes, and it is the most common path in. The technical skills transfer almost entirely; what has to be built new is scoping under ambiguity, writing evals as acceptance criteria, deploying into environments you do not control, and narrating your reasoning to non-engineers. The strongest application signal is one end-to-end thing you shipped for a real user under real constraints.
Do FDEs write less code than software engineers?
Somewhat less, and different code. Anthropic’s forward-deployed job description describes roughly 40% of the time as building, with 30% customer-facing and 30% feeding learnings back into the product. The code skews toward integrations, retrieval pipelines, agents, evals and deployment tooling rather than long-lived product surfaces.
Sources
Every number on this page traces to one of these. Where a figure is self-reported or crowd-sourced rather than first-party, it is labelled inline.
Cited on this page
- Bloomberry — “I analyzed 1,000 forward deployed engineer jobs” — posting growth, builder/pre-sales/internal split. bloomberry.com/blog/i-analyzed-1000-forward-deployed-engineer-jobs-what-i-learned/
- Plank — FDE job-market census — 1,206 strict-definition postings across 669 companies; median posted base $185K. www.joinplank.com/fde-job-market
- Anthropic — Forward Deployed Engineer, Applied AI (JD) — 40/30/30 split; MCP servers, sub-agents and agent skills as deliverables. jobs.menlovc.com/companies/anthropic/jobs/69674588-forward-deployed-engineer-applied-ai
- First Round Review — “So you want to hire a forward deployed engineer” — the staff-engineer bar. review.firstround.com/so-you-want-to-hire-a-forward-deployed-engineer/
- Nabeel S. Qureshi — “Reflections on Palantir” — first-hand account of forward-deployed work, including the Airbus engagement. nabeelqu.co/reflections-on-palantir
- levels.fyi — Palantir FDSE compensation — www.levels.fyi/companies/palantir/salaries/software-engineer/title/fdse
- Perspective — 2026 Forward Deployed Engineering compensation report (1,200 FDEs) — self-reported survey data; treat as medium confidence, not first-party. www.getperspective.ai/blog/2026-forward-deployed-engineering-compensation-report-1200-fdes
Keep comparing
Explore the topic
See this alongside everything else on the same subject — handbooks, system designs, challenges and tools, in one place.
More Handbooks
- The Prompting HandbookA friendly, hands-on field guide for everyday humans — learn the CRISP framework, spot bad prompts, practice with real recipes, play a drag-and-drop game, and test yourself with a quiz. No code required.Read →
- The Agentic AI Interview HandbookTwenty topics every senior AI engineer should be able to reason about live — from eval pipelines to reliability patterns for generative systems.Read →
- The Senior AI Engineer Interview Handbook60 questions across architecture, production incidents, agentic systems, RAG, evals, cost, safety, and leadership — what staff-level AI interviewers actually probe for.Read →
- 50 Angular Interview QuestionsA visual handbook covering components, change detection, RxJS, signals, routing, forms, performance, and testing — what interviewers actually probe for in senior Angular roles.Read →
- 50 Python Interview QuestionsFundamentals to advanced: data structures, OOP, iterators & generators, the GIL, asyncio, memory, testing, and the standard library — a visual walk through everything a Python interview touches.Read →
- 51 LLM Evals Interview QuestionsGolden sets, LLM-as-judge, regression testing, offline vs online evals, RAG evals, agent evals, red-teaming, and observability — demystified for interviews and production.Read →
Explore more from Vibe Engines
Get the next one in your inbox.
New handbooks, system-design walkthroughs, and tools — straight to your inbox. No spam, unsubscribe anytime.