Skip to content

Forward deployed engineer vs AI engineer: what actually differs

The titles overlap, and employers use them differently. Postings from employers that list both give a better test than the title: whose product, whose systems and who decides it worked.

By All Things FDE editorial teamPublished Updated

Contents. Each line links to its section.

Contents

Key points

  • The core split in the postings we read: an AI engineer builds AI features and tooling for the employer's own product, while a forward deployed engineer builds and runs an AI system inside a customer's environment and is judged on whether it works there [1][2][3][4][5].
  • At AI labs and platforms the titles blur: Anthropic's FDE and Applied AI Engineer postings both place the engineer in its Applied AI team, and Scale AI posts the same "Frontier Agents Engineer" title in an Applied AI version and a forward deployed version with the same base salary range [5][6][7][8].
  • Travel and customer contact separate them most clearly in these postings: the FDE postings we read list travel of up to 50% (OpenAI) or 25% to 50% (Anthropic, Glean), while Glean's in-house AI engineer posting lists no travel and no customer-facing duties [4][5][9][2].
  • Machine learning depth does not reliably separate them: PyTorch appears in 6% of the FDE postings in one analysis of about 1,000, though Databricks' AI FDE posting asks for years of hands-on data science [10][11].
  • Pay does not follow the title: on October 23, 2026, OpenAI's Enterprise Applied AI Engineer annual salary range ($251,000 to $278,000) sat inside its FDE range ($185,000 to $300,000), and Scale AI listed the same $180,000 to $225,000 annual base range for both versions [12][4][7][8].

In the postings we read, a forward deployed engineer (FDE) builds and runs an AI system inside a customer's environment and is judged on whether it works there [4][5][8]. An AI engineer builds AI features and tooling for their own employer's product [2][3]. In practice the titles overlap: OpenAI, Anthropic and Scale AI each post a customer-facing "Applied AI" job beside their forward deployed job, Databricks puts both "AI Engineer" and "Forward Deployed Engineering" in one title [12][6][7][11], and O*NET lists no occupation with either name [13][14]. So read the posting, not the title, and ask three questions: whose product, whose systems, and who decides it worked.

This page compares live postings from six employers, read on October 23, 2026, with one published analysis of about 1,000 FDE postings. The postings cover different places: Anthropic's FDE posting is for London and Glean's is remote in the US. Each posting is one company's description at one time, so treat the result as a pattern, not a definition. For the role on its own, see what a forward deployed engineer is.

What is the difference between a forward deployed engineer and an AI engineer?

The "AI engineer" side comes first, because the title has been used two ways. Latent Space's 2023 essay on the title describes an AI engineer as a software engineer who builds on foundation models, mainly through APIs, and treats evaluating models, working with tools such as retrieval and agents, and owning product-specific data and evals as the job. It contrasts that with ML engineers and researchers, who handle the foundation models themselves, such as pretraining and general benchmarks [1]. Not every employer uses the title that way. Databricks' AI FDE posting asks for years of hands-on data science with tools such as scikit-learn and PyTorch, and Scale AI's Applied AI posting lists fine-tuning, distillation and reinforcement learning as preferred experience [11][7]. Two postings with the same title can ask for different depth.

Postings for in-house AI engineers fit the shape the essay describes. Glean's AI Engineer, Data and Evals builds the evaluation sets, benchmarks and eval platform for Glean's own AI products, and lists no direct customer-facing duties [2]. Brex's AI Engineer, Workbench builds the agent workbench inside Brex's own product, and the job includes talking directly to customers and iterating with them [3].

Postings for FDEs describe a different shape. OpenAI's has FDEs own discovery, technical scoping, system design, build and production rollout alongside strategic customers, with success measured by production adoption and measurable workflow impact [4]. Anthropic's London posting has them build production applications with Claude inside customer systems [5].

The shortest test is ownership. In the in-house postings, the AI engineer is accountable for the quality of an AI product the company offers to many customers [2][3]. In the FDE postings, the engineer is accountable for the result at each customer they work with, even when that also means feeding what they learn back into the company's product [4][5]. Neither title maps to a standard occupation: a keyword search of O*NET OnLine, the U.S. Department of Labor's occupation database, returns no occupation titled AI Engineer or Forward Deployed Engineer, so employers define both roles themselves [13][14].

How do the two roles compare side by side?

The table has three columns because the postings show a third role between the other two. An "Applied AI Engineer" at a model vendor sits between them. Anthropic's posting is customer-facing but advisory, while OpenAI's Enterprise posting describes designing, building and deploying systems with customers, much like its FDE posting [6][12].

Table 1. How do the two roles compare side by side?
 AI engineer (in-house)Applied AI engineer (model vendor)Forward deployed engineer
Whose productThe employer's own, such as Glean's AI products or Brex's agent workbench [2][3]Customers' products built on the vendor's models, such as companies adopting the Claude API [6]A system for a specific customer's problem, such as Claude applications built inside customer systems [5]
Customer contactNone listed at Glean; direct customer feedback on the product at Brex [2][3]Direct partnership with customer teams; Anthropic's posting calls the role a technical advisor [12][6]Embedded with customer teams [4][5]
Where the work runsThe company's own product and platform, including eval and simulation environments at Glean [2][3]At Anthropic, customized pilots, prototypes and evaluation suites, with workshops and code reviews for customer engineering teams; at OpenAI, systems designed, built and deployed with customers [6][12]Inside the customer's systems and enterprise environments [5][8]
Success measured byQuality of the AI product, such as evals that define what good looks like [2]Production systems, sustained adoption and customer impact, at OpenAI [12]Production adoption and measurable workflow impact, at OpenAI [4]
Posted experience bar5+ years of software engineering at Glean; 5+ years of software development at Brex [2][3]4+ years in technical roles at Anthropic, which names FDE as one of them [6]4+ years at Anthropic, Scale AI and Glean; 5+ years at OpenAI [5][8][9][4]
TravelNone stated at Glean or Brex [2][3]Occasional at Anthropic; none stated at OpenAI [6][12]Up to 50% at OpenAI; 25% to 50% at Anthropic and Glean; once every 4 to 8 weeks, as needed, at Databricks [4][5][9][11]

Read the table as a spread, not three boxes. The middle and right columns collapse into each other at some employers, which the next section shows. If the job you are weighing is closer to pre-sales demos than production builds, forward deployed engineer vs solutions engineer covers that boundary.

Why do the titles overlap so much?

Employers that post both titles show it. Four examples from the postings we read:

  • Anthropic. The FDE posting and the Applied AI Engineer posting both open by placing the engineer in the Applied AI team. The FDE (London posting) builds production applications inside customer systems and travels 25% to 50%. The Applied AI Engineer advises customers on architecture, builds pilots and evaluation frameworks, and travels occasionally. The Applied AI Engineer posting names Forward Deployed Engineer as one of the technical roles its 4+ years of experience can come from [5][6].
  • OpenAI. Both the FDE and the Enterprise Applied AI Engineer postings describe taking a customer from use case to a production system and judge success by production adoption. The FDE posting is explicit about owning discovery through rollout, asks for 5+ years and lists travel up to 50%. The Applied AI Engineer posting names forward-deployed engineering among its relevant backgrounds and states no travel requirement [4][12].
  • Scale AI. The two postings share a title, "Frontier Agents Engineer", and differ in the bracket and in emphasis. The Applied AI version stresses experiments, evaluation design and machine learning fundamentals. The forward deployed version stresses distributed systems, cloud infrastructure and integrations. Both say every engineer builds skill in both areas over time [7][8].
  • Databricks. One posting is titled "AI Engineer - Forward Deployed Engineering (AI FDE)". It is customer-facing, owns production rollouts of generative AI applications and asks for willingness to travel once every 4 to 8 weeks, as needed [11].

The pattern: at these companies, "Applied AI" and "forward deployed" read as two points on one customer-facing spectrum. In these postings the title tells you less than the responsibilities, travel and measure of success do.

Does a forward deployed engineer need machine learning depth?

It depends on the employer. Bloomberry's analysis of about 1,000 FDE postings, supplied by a jobs-data provider and published in November 2025, found Python in 66% of postings, PyTorch in 6% and TensorFlow in 4%. Working directly with customers was the most common responsibility, at 55% [10]. The author sorted the postings into three types and says the article focuses on the builder type (60% of postings), without saying whether every percentage covers all postings or only that type. The sample also comes from one provider, so it is a rough guide, not a census.

Individual employers differ. Scale AI's forward deployed version asks for distributed-systems fundamentals and experience building or deploying applications with LLM APIs, agent frameworks and retrieval, while its Applied AI version requires a strong understanding of machine learning fundamentals [8][7]. Databricks asks its AI FDE for extensive hands-on data science experience [11]. The in-house AI engineer is not automatically the more mathematical role either: Latent Space argued that shipping AI products takes software engineers more than researchers [1].

Practically, check whether the posting lists model training and evaluation methods as requirements or software delivery and integration, and prepare for the one it asks for.

Which pays more, a forward deployed engineer or an AI engineer?

This page uses posted ranges instead of a salary dataset, because a posting names the employer, the date and sometimes whether the range is base pay. The cleanest evidence is employers that post both roles, with annual ranges listed on October 23, 2026. Each range is one posting at one time and may span several levels, so none of these figures is what a person is paid.

  • OpenAI. FDE (San Francisco) salary range $185,000 to $300,000; Enterprise Applied AI Engineer (San Francisco or New York, one pay zone listed) $251,000 to $278,000. Both postings say equity is offered [4][12].
  • Scale AI (San Francisco, New York, Seattle). The same base salary range of $180,000 to $225,000 for the Applied AI and forward deployed versions. Both postings say eligible roles are also granted equity, subject to board approval [7][8].
  • Glean. AI Engineer, Data and Evals: standard base salary range $200,000 to $280,000. Founding Forward Deployed Engineer: "standard range" of $160,000 to $270,000, which the posting does not label as base. The posting says the FDE range spans several levels. Both say certain roles may be eligible for variable compensation or equity [2][9].
  • Databricks. The AI FDE posting (all levels) lists a local pay range of $182,000 to $250,208, described as the expected salary range for non-commissionable roles. The posting says the total package may also include an annual performance bonus and equity [11].

What this shows: in these postings the ranges overlap heavily and the title does not predict which is higher. At OpenAI the FDE range is much wider. At Scale AI the two are identical. At Glean the in-house AI engineer range starts $40,000 higher. Compare ranges for the same employer, city and level, and ask whether a figure is base or total before comparing it with another.

Which role should you choose, and can you move between them?

The postings suggest three practical tests.

  • Who judges the work. If you want to be judged on a customer's outcome, in systems you do not control, FDE and customer-facing Applied AI roles fit. If you want to be judged on the quality of a product many customers use, in-house AI engineering fits [4][2].
  • Travel. The FDE postings we read list 25% to 50% travel at Anthropic and Glean and up to 50% at OpenAI. The in-house postings we read state no travel requirement [5][9][4][2][3].
  • Customer problems or product quality. Glean's founding FDE posting describes finding unsolved business problems at its most strategic customers and building new products to solve them, the best of which ship to every Glean customer, while its AI engineer posting describes depth in data and evaluation science for Glean's AI products [9][2].

On switching, some employers treat the roles as adjacent. Anthropic's Applied AI Engineer posting names Forward Deployed Engineer, Product Engineer, Software Engineer and Technical Product Manager among the technical roles its 4+ years of experience can come from, and OpenAI's Enterprise posting names applied AI or ML engineering and forward-deployed engineering among its relevant backgrounds [6][12]. Both postings stress production engineering. What each side must add is different: customer-facing delivery experience going toward FDE, which OpenAI's posting asks for explicitly, and evaluation and benchmark-building depth going toward in-house AI engineering, which Glean's posting asks for [4][2]. For the fuller picture of the FDE side, see what a forward deployed engineer is.

What this comparison cannot tell you. We read postings from six employers, mostly AI labs and AI platform companies, plus one fintech. Glean's FDE posting is for a founding role focused on building new products, so it may differ from other FDE postings. Banks, consultancies and staffing firms may use the same titles differently. Postings show what employers ask for, not how often people move between the roles or what people earn.

Common questions

Is a forward deployed engineer the same as an AI engineer?

No, but the titles overlap. In the postings we read, an in-house AI engineer builds and evaluates AI features for the employer’s own product, while a forward deployed engineer builds and runs an AI system inside a customer’s environment and is judged on the result there. At Anthropic and Scale AI, customer-facing Applied AI jobs and FDE jobs share a team or a title and overlap in many duties, so judge the responsibilities, travel and measure of success rather than the title.

Is an applied AI engineer the same as a forward deployed engineer?

Sometimes. Anthropic’s FDE and Applied AI Engineer postings both place the engineer in its Applied AI team, and Scale AI’s two versions of its Frontier Agents Engineer posting share one base salary range. The differences in those postings are emphasis: Applied AI roles lean toward advising, experiments and evaluation, forward deployed roles toward building and operating inside the customer’s systems, and the FDE postings at Anthropic and OpenAI list more travel. At other employers the titles may mean different things.

Do forward deployed engineers need a machine learning background?

Not always. In one analysis of about 1,000 FDE postings, Python appeared in 66% and PyTorch in 6%, and the most common responsibility was working directly with customers. Some employers ask for more: Databricks’ AI FDE posting expects extensive hands-on data science experience. Strong software engineering, LLM application experience and customer communication are the more consistent requirements in the postings we read.

Which pays more, a forward deployed engineer or an AI engineer?

Neither consistently. On October 23, 2026, OpenAI’s Enterprise Applied AI Engineer annual salary range sat inside its FDE range, Scale AI listed the same annual base range for both versions of one job title, and Glean’s in-house AI engineer base range started $40,000 above its FDE range, although only the AI engineer figure is labeled base. Posted ranges can span several levels and are not what people are paid, and some postings do not say whether a range is base or total.

Can an AI engineer become a forward deployed engineer, or the other way around?

Some employers describe the roles as adjacent. Anthropic’s Applied AI Engineer posting names Forward Deployed Engineer among the technical roles its experience requirement can come from, and OpenAI’s Enterprise posting names both applied AI or ML engineering and forward-deployed engineering among its relevant backgrounds. This page has no data on how often people move, so treat confident claims about how easy it is with caution. In these postings, the gap to close is customer-facing delivery experience going one way and evaluation depth going the other.

Sources

  1. [1]Latent Space, “The Rise of the AI Engineer”, 30 Jun 2023, accessed 23 Oct 2026. Back to text
  2. [2]Glean, “AI Engineer - Data and Evals”, 22 Oct 2026, accessed 23 Oct 2026. Back to text
  3. [3]Brex, “AI Engineer, Workbench”, 1 Oct 2026, accessed 23 Oct 2026. Back to text
  4. [4]OpenAI, “Forward Deployed Engineer (FDE) - SF”, 11 Sep 2026, accessed 23 Oct 2026. Back to text
  5. [5]Anthropic, “Forward Deployed Engineer”, 14 Sep 2026, accessed 23 Oct 2026. Back to text
  6. [6]Anthropic, “Applied AI Engineer, Enterprise”, 20 Oct 2026, accessed 23 Oct 2026. Back to text
  7. [7]Scale AI, “Frontier Agents Engineer (Applied AI)”, 22 Oct 2026, accessed 23 Oct 2026. Back to text
  8. [8]Scale AI, “Frontier Agents Engineer (Forward Deployed Engineering)”, 22 Oct 2026, accessed 23 Oct 2026. Back to text
  9. [9]Glean, “Founding Forward Deployed Engineer”, 20 Oct 2026, accessed 23 Oct 2026. Back to text
  10. [10]Bloomberry, “What I learned analyzing 1K forward deployed engineer jobs”, 18 Nov 2025, accessed 23 Oct 2026. Back to text
  11. [11]Databricks, “AI Engineer - Forward Deployed Engineering (AI FDE)”, 20 Oct 2026, accessed 23 Oct 2026. Back to text
  12. [12]OpenAI, “Applied AI Engineer, Enterprise”, 8 Jul 2026, accessed 23 Oct 2026. Back to text
  13. [13]O*NET OnLine, National Center for O*NET Development (U.S. Department of Labor), “O*NET OnLine Occupation Keyword Search: artificial intelligence engineer”, accessed 23 Oct 2026. Back to text
  14. [14]O*NET OnLine, National Center for O*NET Development (U.S. Department of Labor), “O*NET OnLine Occupation Keyword Search: forward deployed engineer”, accessed 23 Oct 2026. Back to text

How this entry is made: each factual claim carries a numbered note to one of the sources above, and the date after each source is when the All Things FDE editorial team last checked it. Editorial standards.