What is a Forward Deployed Engineer? the AI Role Everyone is Suddenly Hiring

A forward deployed engineer (FDE) is a software engineer who embeds inside a customer's business — often for months — and builds production systems on that customer's real data, real workflows, and real constraints. Not demos. Not recommendations. Working software, deployed where the work happens, with the engineer staying through go-live until the system reliably delivers value.

If you've been anywhere near enterprise AI in the last year, you've heard the title. Job postings for forward deployed engineers grew more than 1,100% year over year in 2025, and in a single month this spring, OpenAI and Anthropic committed a combined $5.5 billion to ventures built around the role. This post explains what the job actually is, where it came from, why it exploded, and — the part that matters if you run a mid-market company — how to get FDE-style help when the AI labs' own teams are reserved for accounts far larger than yours.

Where the Role Came From

Palantir invented the forward deployed engineer roughly two decades ago. Its intelligence-agency customers couldn't articulate requirements through normal product discovery — the problems were classified, messy, and poorly understood even by the people living them. So Palantir stopped asking and started embedding: engineers sat inside the customer's environment, learned how the work actually flowed, and built against reality instead of a spec.

Internally, Palantir calls these engineers "Deltas," as distinct from "Devs" who build the core product — and until 2016, Palantir employed more Deltas than product engineers. Shyam Sankar, Palantir's thirteenth employee and now its CTO, is credited as the first FDE, and his description of the job is still the best one: the forward deployed engineer absorbs pain and excretes product. The engineer eats the customer's mess — undocumented workflows, dirty data, systems that don't talk — and what comes out the other side is software that works, plus lessons that flow back into the vendor's core product.

That last part is the quiet genius of the model. A traditional software engineer builds one capability for many customers. An FDE enables many capabilities for a single customer — and the patterns they discover in the field become the product roadmap.

Why the Role Exploded in the AI Era

For fifteen years, the FDE was a Palantir curiosity. Then large language models got good, and every AI company discovered the same thing Palantir learned in 2006: the model is not the bottleneck. Deployment is.

Enterprise buyers of AI have working demos and stalled rollouts. The gap between "the model can do this" and "this runs in production against our systems" is exactly the gap FDEs were invented to close. So the AI industry copied the playbook, at scale:

• Job postings titled "forward deployed engineer" grew over 800% between January and September 2025, and a Bloomberry analysis of more than 1,000 postings put the year-over-year increase at 1,165%.
• In May 2026, OpenAI launched the OpenAI Deployment Company, a joint venture with more than $4 billion in committed capital led by TPG, built specifically to field forward deployed engineers at enterprise customers.
• The same month, Anthropic announced an enterprise AI services company with Blackstone, Hellman & Friedman, and Goldman Sachs, with roughly $1.5 billion committed — explicitly aimed at mid-sized companies like community banks, regional health systems, and manufacturers.
• Venture capital noticed too. Andreessen Horowitz called the FDE the hottest job in startups and described the strategy plainly: AI startups are deliberately trading software margin for a deployment moat, because the company that gets AI working inside a customer's operations is nearly impossible to displace.

When the two largest AI labs put a combined $5.5 billion behind a job title in one month, the title stops being a curiosity. It's a signal about where the actual value in AI now lives: not in model access, but in deployment.

What an FDE Actually Does

Strip away the mystique and the job has a recognizable arc:

Map how the business actually works. Not the org chart version — the real version. Where data lives, which workflows exist only in one employee's head, which "system of record" is actually a spreadsheet, where exceptions pile up. This mapping phase is most of the value, because no vendor demo survives contact with a company's real data.

Build and integrate in production. FDEs write production code inside the customer's environment: wiring the model into the CRM, the dispatch system, the document pipeline, the ticketing queue. In the Bloomberry postings analysis, the most common responsibilities were working directly with customers, building and deploying AI systems, and integrating APIs — engineering work, done on-site or embedded remotely.

Stay through adoption. This is what separates the role from a consulting engagement. The FDE is measured on whether the system keeps running and keeps getting used after go-live — tuning reliability, handling the edge cases that only show up in week three, and driving usage until the thing is load-bearing.

FDE vs. Solutions Engineer vs. Consultant

The title gets confused with several adjacent roles, so here's the clean separation:

Solutions engineers and sales engineers work pre-sale. They build demos and proofs of concept to win the deal, and their involvement winds down at signature. They're optimized for deal velocity. An FDE starts roughly where the SE stops — after the contract, when someone has to make the promise true in production.

Consultants deliver recommendations; FDEs deliver running systems. A traditional consulting engagement is judged at handoff: the deck, the roadmap, the configuration. An FDE is judged months later, on whether the software is still running and still creating value. It's the difference between advice and outcomes.

Implementation engineers configure; FDEs build. Implementation teams set up existing product features. FDEs write new code to close the gap between what the product does and what this customer's operations require — and feed what they build back to the product team.

What Forward Deployed Engineers Cost

Across more than 1,000 disclosed job postings, the median FDE salary was roughly $174,000. That's the broad market. At the top of it, Palantir's forward deployed software engineers earn a reported $171,000 to $295,000 in total compensation, and the frontier labs pay multiples of that — reported total compensation for mid-to-senior FDEs at OpenAI runs $350,000 to $550,000, with senior levels reported above $1 million.

Those numbers explain two things at once: why the role attracts strong engineers, and why almost no mid-market company will ever employ one. A single FDE at median market comp is a larger line item than most mid-market companies' entire software budget — before you account for the fact that one engineer, alone, without a product organization behind them, is not actually the model.

The Mid-Market Gap

Here is the part the hiring statistics don't say out loud: vendor FDE teams go where the contracts are biggest. OpenAI's forward deployed engagements target problems worth tens of millions of dollars and up. Palantir's model has always assumed government and Fortune 500 budgets. Even the Anthropic joint venture — the one explicitly aimed at mid-sized companies — is a multi-billion-dollar acknowledgment that this tier has been unserved. As Goldman Sachs' Marc Nachmann put it, the venture exists to "democratize access to forward-deployed engineers" for companies that can't afford the talent or the consulting fees.

But a mid-market operator's problems are structurally identical to the enterprise version: data scattered across systems that don't talk, workflows that live in people's heads, AI pilots that demo well and die in production. The problems didn't get smaller. The access did.

This is, candidly, the gap Queen City AI was built for — and why we've worked forward deployed since before we knew the title was about to become the industry's hottest job. We do discovery before we build anything, because mapping how your business actually runs is most of the work. We build inside your systems, not next to them. And we stay through production, because proof beats promise — a system that's still running in month six is the only deliverable that matters. If you're evaluating outside help, our buyer's checklist for choosing an AI consultant is written to make that distinction easy to test in a first conversation, whoever you talk to.

You don't need to hire a forward deployed engineer. You need the model: embedded engineering, production code, and accountability that outlasts the kickoff deck.

Frequently Asked Questions

What does a forward deployed engineer actually do?

A forward deployed engineer embeds inside a customer's business, maps how work and data actually flow, writes production code to integrate AI into daily operations, and stays through go-live until the system reliably delivers value. Unlike a consultant judged on deliverables at handoff, an FDE is measured on whether the software keeps running and keeps being used.

Where did the forward deployed engineer role come from?

Palantir invented the role roughly two decades ago for intelligence-agency customers who couldn't spec their own requirements. Internally called "Deltas," these embedded engineers outnumbered Palantir's product engineers until 2016. Shyam Sankar, Palantir's thirteenth employee and current CTO, is credited as the first FDE.

What is the difference between a forward deployed engineer and a solutions engineer?

A solutions engineer works pre-sale: demos, proofs of concept, and technical support to win the deal, ending at signature. A forward deployed engineer works post-sale: production code inside the customer's environment, with ongoing ownership of the deployment. Solutions engineers optimize for deal velocity; FDEs optimize for deployment success.

How much does a forward deployed engineer make?

Across more than 1,000 disclosed postings in 2025, the median forward deployed engineer salary was about $174,000. Palantir FDSEs earn a reported $171,000–$295,000 in total compensation, and frontier labs pay more — reported $350,000–$550,000 for mid-to-senior FDEs at OpenAI, with senior levels reported above $1 million.

Why are OpenAI and Anthropic hiring so many forward deployed engineers?

Because model quality is no longer the bottleneck — deployment is. FDE job postings grew 800–1,165% in 2025, and in May 2026 both labs launched dedicated deployment ventures: the OpenAI Deployment Company with over $4 billion committed, and Anthropic's enterprise services company with Blackstone, Hellman & Friedman, and Goldman Sachs at roughly $1.5 billion.

Does my company need a forward deployed engineer?

You need FDE-style help if you're past the demo stage and stuck on integration: data in disconnected systems, workflows that exist only in employees' heads, or AI pilots that never reached production. Most mid-market companies shouldn't hire one at $174,000-plus median compensation — the practical question is how to get the embedded-engineering model, not the headcount.

How do mid-market companies get forward deployed engineering without hiring one?

Vendor FDE teams concentrate on the largest accounts, which is why even Goldman Sachs describes the new Anthropic venture as an effort to "democratize access to forward-deployed engineers." Mid-market companies typically get the same model through AI consultancies that work embedded — doing discovery inside your operations, building in production against your real systems, and staying accountable after go-live.

Related Reading

• Why We Do Discovery Before We Build Anything — The first half of the FDE model, in practice: mapping how your business actually runs before writing a line of code.
• How to Choose an AI Consultant: A Buyer's Checklist — The questions that separate embedded, production-grade partners from deck-and-dash consulting.
• From AI Interest to AI Advantage: Why Proof Beats Promise — Why a system running in production is the only AI deliverable that matters.

Related Solutions

• Agent Systems — AI agents built inside your workflows, deployed to production, and supported after go-live.
• AI Game Plan — The discovery and roadmap phase of forward deployed work: find where AI pays back before you build.