The short answer
- What it is: a forward deployed engineer (FDE) is a software engineer who works inside one customer's workflow, systems and data, builds the working system there, and feeds the patterns that repeat back into the product. Palantir's shorthand is "one customer, many capabilities".
- Where it comes from: Palantir, which says it "pioneered" the Forward Deployed Software Engineer role and calls these engineers Deltas.
- Why now: Anthropic launched an AI services company on 4 May 2026 and OpenAI launched its Deployment Company on 11 May 2026. Lightcast data shows FDE postings up more than 1,000% for January to August 2026 against the same months of 2025.
- What the job asks: production code, hands-on work with models or data, judgement in front of the customer, and posted travel of 20% to 50% depending on the employer.
- In Switzerland: on 30 September 2026, Google lists four roles with Forward Deployed Engineer in the title that include Zurich, and Databricks lists a senior one. OpenAI, Anthropic and Palantir list none there.
May 2026: two AI labs built companies around one job title
Within seven days in May 2026, Anthropic and OpenAI each launched a separate company built around engineers who work inside customer organisations. If you write code in Switzerland, a forward deployed engineer posting is already open in Zurich. If you run a team, both labs now offer this way of working through their new companies.
Two launches, one week apart
On 4 May 2026, Anthropic announced a new AI services company for mid-sized firms, together with Blackstone, Hellman & Friedman and Goldman Sachs. Anthropic's Applied AI engineers work alongside the new company's engineers, and an engagement starts with "a small team working closely with the customer". Anthropic's CFO Krishna Rao gave the reason: "Enterprise demand for Claude is significantly outpacing any single delivery model."
One week later, on 11 May, OpenAI launched the OpenAI Deployment Company to embed Forward Deployed Engineers in organisations. It starts with more than USD 4 billion in initial investment and 19 partners, and it is acquiring Tomoro, which brings about 150 FDEs and Deployment Specialists.
TechCrunch reported a valuation of USD 1.5 billion for the Anthropic venture and tied both ventures to "the forward-deployed engineer (FDE) model popularized by Palantir." Two labs that compete on models chose the same delivery model within a week.
What the posting data shows
Lightcast data reported by Fortune on 3 September 2026 shows FDE postings up more than 1,000% for January to August 2026 against the same period of 2025, and more than 4,600% against 2023. Broader tech postings rose 13%.
A rise of more than 1,000% means this year's volume is over eleven times the 2025 level, so the base was small. For you as an applicant in Switzerland, Google and Databricks both list FDE roles in Zurich.
What a forward deployed engineer is
One customer, many capabilities
Palantir's student careers page splits its engineers in two. Devs sit in Product Development and work on "one capability, many customers". Deltas, the forward deployed software engineers, sit in Business Development and work the other way round: "one customer, many capabilities".
From that split, I define an FDE as a software engineer whose unit of work is one customer. The code runs inside that customer's systems and data, and the patterns that repeat travel back to the product team.
Where the title comes from
Palantir's own FDSE posting says the company "pioneered" the role and describes it as embedding engineers directly with its customers. On the Palantir board, 81 of 318 postings on 29 September 2026 still have "Forward Deployed" in the title.
The AI labs now reuse the title for engineers who deploy their models inside customer systems. OpenAI's job board has a whole Forward Deployed Engineering department, and Anthropic hires FDEs in Europe and the US.
How it differs from software and solutions engineering
A software engineer on a product team, Palantir's Dev, builds one capability that many customers use. A solutions engineer is, in the words of OpenAI's Solutions Engineer posting, "the technical expert in the sales process", with pre-sales work, demos and proofs of concept.
The commercial difference shows in the Databricks posting: "FDEs are billable." A solutions engineer helps win the deal. At Databricks, an FDE's time is part of what the customer pays for.
What the work looks like day to day
Discovery inside the customer's workflow
In OpenAI's FDE posting, one role owns "discovery, technical scoping, system design, build, and production rollout". That puts one engineer on the work from the first conversation to the live system.
Building in systems you do not own
Anthropic's FDE posting says its FDEs "work within customer systems" and deliver "MCP servers, sub-agents, and agent skills". MCP servers are connectors that give a model access to a tool or a data source. Palantir's careers page says "Deltas build" and lists work such as "delivering scalable data infrastructure".
OpenAI's Deployment Company describes the same thing from the customer side: FDEs build inside the organisation, connecting models to the customer's data, tools and controls. Much of the job is wiring a model into systems that already exist, with access rules and data you did not design.
Feeding what repeats back into the product
In my reading, this half of the job is what separates an FDE from a contractor. OpenAI asks its FDEs to "codify working patterns into tools, playbooks, or building blocks" and to share field feedback with Research and Product. Anthropic asks for the same in its own words: "codify repeatable deployment patterns and contribute insights back".
Palantir ties it to the platform itself. Its Deltas work on "expanding Palantir's core platform to solve new problems as they are discovered." So an FDE's findings can change what the next customer gets.
For an applicant, this is the part to probe in an interview. If the role has no route back into the product, I would treat it as a consulting job with a new title.
Travel and time on site
Posted travel differs by employer:
- Databricks: 20%
- Anthropic: about 25% in its US posting, 25% to 50% in Paris, Munich and London
- Palantir: up to 25%
- OpenAI: up to 50%
The skills the job asks for
Production code
Every posting checked asks for engineers who ship code, and the experience bar varies widely. Palantir's FDSE posting asks for 1+ year of experience and strong coding in a language such as Python, Java, C++ or TypeScript. Anthropic asks for strong Python and 4+ years in a technical, customer-facing role in its US and London postings, and 8+ years in Paris and Munich. OpenAI asks for 5+ years, and the senior Databricks role in Zurich for 6+ years in data engineering, data platforms or software engineering, plus deep experience with Apache Spark.
Hands-on experience with models
Anthropic asks for production LLM experience, including evaluation, meaning systematic tests of a model's output against expected results. Palantir's requirements, listed above, centre on coding instead. How much model work you do depends on the employer.
Judgement in front of the customer
OpenAI's posting lists discovery and technical scoping before system design and build, so the engineer is in conversations with the customer before any code is written.
As I read the postings, the code is the part you can prepare for. The harder part is deciding, with people who do not write code, what gets built first and what stays out.
The forward deployed data engineer
Among the postings checked, the Databricks role in Zurich comes closest to a forward deployed data engineer. Its title says Fullstack, but it covers data engineering, AI and application development, and asks for Apache Spark (an open-source engine for processing large datasets), 6+ years of experience and 20% travel.
Forward deployed engineer jobs in Zurich and Switzerland
What is open on 30 September 2026
For this article I checked the boards of Google, Databricks, OpenAI, Anthropic and Palantir. Two employers on the boards checked list the title in Zurich. Google's careers site shows four roles with Forward Deployed Engineer in the title that include Zurich, among them a Senior Forward Deployed Engineer, GenAI, Google Cloud based there. Databricks lists a Sr. Forward Deployed Engineer (Fullstack) in Zurich, first published on 16 September 2026.
The others checked on the same day list nothing in Zurich. OpenAI's job board shows 26 roles in its Forward Deployed Engineering department, none in Zurich. Anthropic's board lists FDE roles in Paris, Munich, London and the US. Palantir's board has 81 postings with "Forward Deployed" in the title and none in Switzerland.
If you want the title without leaving Switzerland, Google and Databricks are the two employers on these boards that have it.
What the pay data covers
The postings checked that state pay are in the US, London and Paris:
| Employer, location | Posted pay range |
|---|---|
| Palantir, New York | USD 135,000 to 200,000 |
| OpenAI, San Francisco | USD 185,000 to 300,000 plus equity |
| Anthropic, US | USD 280,000 to 320,000 |
| Anthropic, London | GBP 225,000 to 255,000 |
| Google, Paris | EUR 104,000 to 106,000 plus a 15% bonus target and equity |
Lightcast puts the median advertised FDE salary above USD 188,000, against about USD 145,000 for software engineers. Fortune's report does not say which countries that median covers, so read it neither as a Swiss figure nor as a global one. No posting checked for this article states pay for Zurich. Google's Forward Deployed Engineer, Applied AI role lists Zurich among its locations but gives a range only for France, and the Databricks role gives none. So there is no CHF figure here, and converting US, UK or French ranges would give you a number no Swiss employer has posted.
What it means for a company that brings one in
An engagement starts small
Both May 2026 announcements describe a similar start. Anthropic's services company begins with "a small team working closely with the customer". OpenAI's Deployment Company begins with a diagnostic, then "a small number of priority workflows".
What you have to provide
Access to the real systems and data comes first. Both labs' FDEs work inside the customer's systems, and neither can do that without logins and someone on your side who can grant them.
You also provide the people who know where the time goes. The engineer builds around their knowledge of the workflow, so their hours are part of the cost. Before you sign, ask who on your side will spend time with the engineer each week. If nobody can, the build rests on guesses.
What stays when the engineer leaves
What stays is a production system connected to your data and controls. Read the "codify" lines in the postings from the customer side, though: the patterns an FDE writes down go into the vendor's tools and playbooks.
I think that is fair, and it is how the product improves. It also means you should agree before the work starts which code, documentation and access remain yours when the engagement ends.
The two halves of the job in my own work
I have not worked as an FDE for a client. My own blog tooling still showed me both halves of the job in its first three days.
A build for one team became an engine
The blog automation I had built for one team is now one engine. Each of my sites is a config file plus a facts record, a list of every claim an article may make about my work. I ported the method, never the team's content.
The first engine commit landed on 27 September 2026, and the first article through it went live on the seelig.ai blog on 29 September. So far, seelig.ai is where its output has appeared. That is the codify half: what repeated in one team's build became the engine, and the team's own material stayed behind.
The real environment broke an assumption
On the first run, a markdown table passed the self-check, the fact gate and the copy review. Only the review that compiled the real page caught it, because my site renders MDX without table support. Every check that read the text alone was blind to the environment. That is the half of forward deployed work that only shows inside the real system.
A lint rule now blocks tables on that site. A second guard follows the same idea: a prompt with a required slot left empty refuses to run instead of running with a gap.
If you want an engineer inside your own workflow
If a vendor has pitched you forward deployed engineers and you want to see what that looks like at your scale, building an AI system inside a team's own workflow is work I take on. Bring one workflow to a first conversation, ideally the one where your team copies the same data between systems every week.
Frequently asked questions
Is a forward deployed engineer a consultant with a different title?
Not in the postings checked. Palantir, OpenAI and Anthropic all describe an engineer who writes and ships production code inside the customer's systems. The same employers ask FDEs to feed the patterns they find back to a product team, and that link to a product is what the title adds.
Do I need AI or LLM experience to apply?
It depends on the employer. Anthropic asks for production LLM experience, including evaluation. Palantir's FDSE posting asks instead for 1+ year of experience and strong coding in a language such as Python, Java, C++ or TypeScript.
How much does a forward deployed engineer earn?
The postings checked that state pay are in the US, London and Paris. Palantir's New York range is USD 135,000 to 200,000, OpenAI's San Francisco range USD 185,000 to 300,000 plus equity, Anthropic's US range USD 280,000 to 320,000, Anthropic's London range GBP 225,000 to 255,000 and Google's range for France EUR 104,000 to 106,000 plus a 15% bonus target. None of the postings checked states a salary for Zurich.
Are there forward deployed engineer jobs in Zurich?
Yes. On the boards checked on 30 September 2026, Google lists four roles with Forward Deployed Engineer in the title that include Zurich, and Databricks lists a senior FDE role published on 16 September. OpenAI and Anthropic list none in Zurich, and Palantir none in Switzerland. Anthropic's European FDE roles sit in Paris, Munich and London.
What is a forward deployed data engineer?
An FDE whose work centres on data. The Databricks role in Zurich comes closest among the postings checked: data engineering, AI and application development, Apache Spark and 6+ years of experience, with the engineer billable to the customer.