What Is a Forward Deployed Engineer?

In the spring of 2026, OpenAI and Anthropic both announced billion-dollar joint ventures organized around the specific task of deploying AI inside enterprise customers. OpenAI raised $4 billion for the effort. Anthropic raised $1.5 billion. Both ventures are built around a role called the Forward Deployed Engineer (FDE), a title that barely existed five years ago and now represents some of the fastest-growing hiring demand in the industry.

I've been watching this from the NeuronHire intake desk for the past year. In the first half of 2026, roughly one in five companies that came to us with an engineering need described something that didn't fit the standard software engineer profile: someone who could work directly inside a customer's infrastructure, own a deployment outcome end-to-end, and operate without a complete specification. Most of them didn't have a name for it. Well, they were looking for an FDE. They just didn't know it yet.

What exactly is a Forward Deployed Engineer?

An FDE is a software engineer who works inside a customer's production environment, writing code against the customer's actual systems rather than an internal codebase or controlled sandbox. The job is finished when the customer's team is using the system in production.

The structural difference from consulting or solutions engineering is what happens to the work afterward. A consultant delivers a scoped project and moves on; the output belongs to one client. In the FDE model, at least in its original form, solutions built for one customer get generalized and folded back into the core product. Palantir (which kind of invented this role) called this the "gravel road to paved highway" approach. The FDE builds the custom gravel road; if enough customers need it, engineering paves it into the platform.

That feedback loop is also the thing worth examining closely before accepting or offering any FDE role. Without it, the model is professional services with a more appealing job title.

Where did the Forward Deployed Engineer role come from?

Palantir invented it in the early 2010s, driven by a specific constraint: their clients were US intelligence agencies that couldn't clearly describe what they needed and couldn't share data with outside teams. Standard product discovery methods were off the table. The only way to build software that worked was to embed engineers inside the client environment and learn the requirements by building.

Palantir organized this around a two-person model: Echo and Delta. The Echo brought domain expertise in the client's industry. The Delta brought rapid prototyping speed. The Delta was the original FDE. Their job was to ship something working and learn from the result, not to design or specify upfront.

By 2016, according to FDE Academy's analysis of the Palantir model, the company had more FDEs than traditional software engineers, a ratio no product company had ever reached. The economics worked because embedded work that proved valuable in one customer's environment was standardized into product features available to all customers. FDE work was a discovery method that happened to run on production infrastructure.

The principle Palantir stumbled onto in classified government environments turned out to describe enterprise software broadly: products fail at the deployment layer, not the product layer.

Why is FDE demand growing 1,165% year over year?

AI products have the same deployment problem that Palantir's government software had in 2012.

A 2024 MIT study found that 95% of enterprise AI projects fail to generate measurable business value, and the main reason is implementation. AI products that work in demo environments fail in enterprise infrastructure because they can't connect to a fifteen-year-old SAP deployment, can't handle a proprietary SSO configuration built by someone who left the company in 2019, and weren't designed for how the people using them actually work.

According to Bloomberry's analysis of more than 1,000 FDE job postings, FDE roles on Indeed grew 1,165% year-over-year from 2024 to 2025. LinkedIn postings grew 800% between January and September 2025. By April 2026, more than 5,300 FDE postings were active on Indeed, up from 643 twelve months earlier. a16z called it "the hottest job in tech."

Companies spent approximately $684 billion on AI in 2025 (per Bloomberry's figures) and largely couldn't show returns on it. FDE capacity is one of the few interventions that addresses the implementation layer directly. That is why OpenAI and Anthropic are organizing billion-dollar ventures around it.

The supply gap compounds the pressure. FDE candidate pools grew approximately 50% year-over-year while postings grew 1,165%. The ratio isn't improving.

How does an FDE differ from a Software Engineer, Solutions Engineer, or Consultant?

The FDE title is now being applied to a wide range of roles, which matters practically to anyone hiring or evaluating an offer. First Round Capital noted that "some companies are clearly rebranding sales engineering jobs as forward deployed engineers." The title alone tells you little.

The clearest structural distinction is deal stage. Solutions engineers work pre-sale: they build proofs of concept, run demos, and help close contracts. Once the deal is signed, their involvement winds down. FDEs start where solutions engineers stop, working inside the customer's real infrastructure rather than a controlled environment, writing code that will run in production after they leave.

The distinction from consulting is less about daily activity and more about where the work goes. Marty Cagan at SVPG makes the point directly: FDEs who embed with multiple customers to find patterns that generalize into product features are a product discovery function. FDEs who serve one enterprise client indefinitely are a professional services team with a different job title.

Bloomberry's dataset confirms one detail that places the role unambiguously: zero percent of FDE job postings in their sample carried a sales quota. Post-sale, technical, outcome-oriented.

Dimension	FDE	Software Engineer	Solutions Engineer
Where they work	Customer's production environment	Internal codebase	Pre-sale sandbox
Code quality	Production-grade, customer-specific	Scalable, reusable	Demo-grade, disposable
Success metric	Live deployment + adoption	Product quality	Deals closed
Quota-bearing?	No	No	Usually yes
Deal stage	Post-sale, implementation	Internal	Pre-sale

What skills does a Forward Deployed Engineer actually need?

Per Bloomberry's skills analysis, Python appears in 66% of FDE job descriptions, TypeScript in 35%, and AWS in 32%. AI agents and LLM experience show up in 35% and 31% of postings respectively; AI deployment is now the primary driver of FDE hiring. Sixty percent of roles target mid-level engineers with three to five years of experience.

The technical requirements are real but not the hardest part of the profile to find. First Round Capital identified five traits that distinguish exceptional FDEs: the habit of solving problems without over-engineering them, the grit to work in genuinely difficult situations, a technical bar equivalent to a staff engineer, a compulsion to ship across multiple contexts, and real curiosity about how organizations function, specifically how people and incentives inside a company work, not just the code.

The T-shaped profile that gets hired consistently: depth in Python or TypeScript plus cloud infrastructure, breadth across systems integration, AI tooling, security, and compliance. The non-technical dimension is what most job descriptions fail to capture: managing ambiguity without a complete requirements document, and translating technical constraints for non-technical stakeholders who are waiting on a decision.

Engineers who thrive in product teams because of their depth and precision in a single stack often find FDE work frustrating. Those strengths are genuinely valuable and genuinely mismatched.

Which companies are hiring FDEs, and what do they pay?

The companies building FDE capacity fastest are the ones whose products land hardest in complex enterprise environments. OpenAI and Anthropic are both scaling FDE teams aggressively following their joint venture announcements. Palantir, which invented the model, continues hiring. Salesforce called FDE "today's hottest role" on its careers site. Ramp, Stripe, Datadog, Cursor, and Cohere all have active FDE functions.

Most of the demand is coming from growth-stage companies rather than large enterprises. These are businesses that have shipped a product complex enough to require embedded engineering support but whose core team can't absorb every customer's integration problem internally, a pattern consistent with what we see in our own intake at NeuronHire. The industries generating the most FDE demand are the ones with the most deployment friction: financial services, government and defense, and healthcare, all heavily regulated sectors where off-the-shelf AI products face the hardest resistance at the implementation layer.

On compensation, Pave's benchmarks show FDEs earning approximately 9.2% below software engineering generalists at comparable seniority levels. At OpenAI and Anthropic, senior FDE total compensation reaches $350,000 to $550,000. FDE Academy's analysis of Palantir's FDE population puts staff-level total comp above $630,000. At earlier-stage companies the numbers run lower, but equity is nearly universal in FDE offers. The market hasn't caught up to how scarce the profile is.

How do you become a Forward Deployed Engineer?

LeetCode preparation, single-stack specialization, and optimizing for internal product teams are not the path. Engineers who land FDE roles typically come from one of two places: solutions engineering, or consulting-adjacent implementation work where direct client contact and incomplete specifications are routine.

The profile that gets hired consistently: three to five years of engineering experience, Python or TypeScript as primary languages, meaningful cloud infrastructure exposure, and at least one role where the outcome required understanding a customer's business rather than just their codebase. Experience integrating systems that weren't designed to communicate with each other is more directly relevant than most certifications.

The non-technical side is harder to develop deliberately. FDE work requires explaining a technical constraint to someone with no engineering background, writing code in situations where requirements arrive late or not at all, and managing the pressure of being the person a customer team is waiting on. Those skills develop in client-facing environments. Seeking out a solutions engineering rotation, an implementation consulting project, or any role with direct external stakeholder contact is a more effective preparation than additional credentials.

Worth knowing before pursuing this path: the burnout rate in FDE roles is higher than in traditional product engineering. Carrying customer expectations and internal engineering standards simultaneously, often without a clear playbook, is demanding in a way that's hard to anticipate. The engineers who sustain FDE careers tend to find genuine energy in customer problems. Those who don't, usually find that out quickly.

How do you hire a Forward Deployed Engineer?

The LeetCode screen is where most FDE searches fail first. It selects for algorithmic depth and performance under a clear specification, which are the wrong signals for a role defined by ambiguity and breadth. Companies running the same technical evaluation they use for core product engineers will systematically pass on their best FDE candidates.

First Round Capital's hiring framework centers on three questions: can this person write production-grade code quickly in a messy environment, can they manage a customer relationship without a script, and can they scope a problem that has never been fully described? A client-scenario case study presenting a realistic integration challenge with incomplete requirements addresses all three in a way that no algorithmic interview can.

Sourcing is the second problem. The FDE profile, T-shaped breadth, client experience, comfort with ambiguity, appears more often in solutions engineering, enterprise implementation, and consulting-adjacent roles than in standard product engineering pipelines. Searching the same candidate pool used for core engineering hires produces the same results it always has.

From our intake work at NeuronHire, Latin American engineers with integration or consulting-adjacent backgrounds carry this profile more frequently than the standard US-market applicant pool reflects. Client-facing communication habits built in environments where you're often the only technical person in the room, combined with T-shaped technical breadth, show up consistently in this population. NeuronHire already places engineers into FDE and FDE-adjacent roles across US companies. The remote model, which companies like PostHog have validated at scale, is where most of our placements sit. If you're looking for engineers who can own deployment outcomes end-to-end, that's the conversation to have.

Should every AI company hire a Forward Deployed Engineer?

No.

FDEs are expensive by design. A staff-level FDE at a frontier AI company costs between $350,000 and $550,000 in annual total compensation. For companies open to the remote model, LATAM engineers with the same FDE profile run between $150,000 and $180,000, based on NeuronHire's placement data. They require real support infrastructure and don't scale the way a product feature does. When the model works, the economics hold up: according to Rocketlane's 2026 FDE guide, one unblocked enterprise implementation typically covers an FDE's annual salary. Intercom scaled Fin AI from 5 design partners to 7,000 customers in 18 months using the FDE deployment model, reaching a 67% AI resolution rate.

When the feedback loop breaks and FDE work never generalizes into product, the result is a consulting practice embedded inside a software company. First Round Capital is direct about this: "Forward deployed engineering is being framed as a panacea right now. But it's a lot more complicated than that." The model works when the product is genuinely difficult to deploy in complex environments and when the company has the discipline to extract patterns from what FDEs find in the field. It doesn't work as a workaround for a product that needs to improve, or as a substitute for making the product easier to deploy.

What should you expect from FDE hiring over the next 18 months?

The title will keep getting diluted. "Full-stack engineer" and "DevOps" went through the same cycle, applied so broadly that they stopped signaling anything specific. Hiring managers who evaluate the title rather than the actual role requirements will make avoidable mistakes.

The underlying demand won't slow down. The structural problem FDEs exist to solve is not going away. OpenAI and Anthropic's joint ventures are multi-year commitments. The absolute volume of FDE hiring is unlikely to compress even if the growth rate moderates.

The supply gap remains the central constraint. FDE candidate pools grew roughly 50% year-over-year while postings grew 1,165%. Companies that find FDE-capable engineers through non-traditional channels, including LATAM engineering markets where the profile characteristics show up more frequently, will have a real hiring advantage over the next two years. The rest will keep searching.

---

NeuronHire connects US companies with Latin American engineering talent. That's our business and our bias.