Picture a VP of AI who posts an opening for an "Applied AI Engineer." The resumes that come back split into two piles: research candidates whose work ended before production, and deployment specialists whose experience began after the product was built. Neither group matches the position, because the company needs someone to own the application layer between them.
Key Takeaways
- Research Engineers improve the underlying model or technical capability.
- Applied AI Engineers turn those capabilities into reusable products or internal workflows.
- Forward Deployed Engineers adapt AI systems to customer environments and own production deployment outcomes.
- Fewer than 2,000 elite FDEs exist within an estimated U.S. market of roughly 17,000, according to Christian & Timbers’ Elite FDE Scarcity Study.
The mistake begins when companies define AI roles by technical skills instead of ownership. Building a model capability requires different experience from turning it into a usable product. Deployment inside a customer environment introduces another level of ownership, centered on production constraints and customer outcomes.
Where ownership actually sits
The clearest distinction is the outcome each engineer owns and how far that ownership extends into production.
A Research Engineer's output is technical: a new training method, an architecture change, an evaluation result that moves a benchmark. They typically report to a Head of Research or VP of AI Research. Applied AI Engineers ship something a team actually uses, a feature or an internal workflow, and get judged on whether it holds up: performance, reliability, adoption. Their reporting line usually runs through a VP of Product or VP of Engineering.
Forward Deployed Engineers work differently. Their output is a system bent to fit one customer's data and constraints, and the only measure that matters is whether that customer's deployment actually worked. Depending on how a company structures deployment, they might report through Engineering, Product, or a dedicated customer-facing technical team.
[Table 1]
How the three roles work together
Organizations that build proprietary AI products and deploy them into enterprise environments may need all three roles. Other companies need only the functions that match their operating model.
Where all three exist, the work moves in a loop. A Research Engineer develops or improves a capability. An Applied AI Engineer turns that capability into a usable product or workflow. A Forward Deployed Engineer adapts the resulting system to a specific customer's environment. The deployment exposes failures and workflow gaps that never appeared in the lab. Those findings return to the product and research teams, shaping the next version.
When a company needs all three functions but leaves one unstaffed, the other teams absorb the cost. A company with strong Research Engineers and no FDEs builds capable models that never survive contact with a real customer environment. A company with FDEs and no Applied AI Engineers ends up with a pile of one-off customer fixes that never turn into a repeatable product.
Which role should you hire?
[Table 2]
Getting the role wrong becomes especially costly when the missing capability is customer deployment. Research and product experience can appear adjacent on a resume, while proven FDE ownership comes from a much smaller market.
A VP of AI should decide which outcome currently lacks an accountable owner. If the technical capability already exists and customer implementations keep stalling, another Research Engineer will not solve the problem. If every deployment becomes a custom build, the missing capability may sit in Applied AI Engineering rather than the FDE team. See our related guide on VP of AI vs. Director of AI for how that reporting structure typically splits.
Why the scarcity data is FDE-specific
Fewer than 2,000 elite Forward Deployed Engineers exist among roughly 17,000 people carrying that title in the U.S., according to our Elite FDE Scarcity Study. Demand for them is projected to grow 2,100% by the end of 2026. One qualifying route to elite status is $10 million or more in documented deployment value. A background at Palantir, Google, Anthropic, OpenAI, or Microsoft in a deployment-specific role is the other. Nearly 80% of the elite FDEs we identified trace back to Palantir.
Microsoft committed $2.5 billion to Microsoft Frontier Company, bringing together 6,000 industry and engineering experts who will work with customers to design, deploy, and improve AI systems. AWS put $1 billion behind a dedicated Forward Deployed Engineering organization that will embed thousands of experts with customers.
Our AI-Native Builder Report found that companies need anywhere from 20 to more than 100 forward deployed engineers depending on deployment scale. The Scarcity Study separately found that only one in 100 of the 180 companies surveyed had reached an AI deployment worth $100 million or more.
These figures describe the FDE market specifically. Comparable research using the same definition and methodology is unavailable for Research Engineers and Applied AI Engineers, so the numbers should not be extended to those roles.
Interview questions that separate real ownership from adjacent experience
Titles on a resume rarely reveal how far someone's ownership extended, so the interview should establish what happened after the technical work left their hands.
What happened after your model or prototype was ready?
A Research Engineer's answer may end with publication or a handoff. An Applied AI Engineer or FDE's answer continues into what happened once real users or a real customer touched it.
Who handled failures once the system entered production?
Their answer reveals how far their ownership extended into production and which team carried operational accountability.
How was your performance measured?
A benchmark score points to Research Engineer work. A shipped feature or adoption metric points to Applied AI Engineer work. A specific customer's revenue or cost outcome points to FDE work.
How closely was your work tied to customer adoption or revenue?
This is the cleanest signal for FDE experience specifically. A vague answer, several steps removed from any customer, suggests that the candidate's experience did not include direct FDE ownership.
FAQ
1. Is an FDE an Applied AI Engineer?
Both roles build with AI. An FDE works inside a specific customer's environment and is measured on that customer's outcome. An Applied AI Engineer's primary responsibility is the reusable product or internal system.
2. Which role works directly with customers?
FDEs have the most consistent direct customer responsibility. Applied AI Engineers can participate in customer conversations, but their primary mandate usually remains the reusable product or internal application.
3. Can one engineer perform all three functions?
At small scale, yes. As deployment volume grows, the three functions usually split into separate roles, because each requires a different kind of judgment under time pressure.
4. Who usually manages each role?
Research Engineers report to a Head of Research or VP of AI Research. Applied AI Engineers report to a VP of Product or VP of Engineering. FDEs commonly report through Deployment, Engineering, Product, or a customer-facing technical organization, depending on whether the company treats deployment as an engineering function or part of its commercial organization.
Before opening the search
Before opening an AI engineering search, define what the hire must own and where that responsibility ends. That decision determines the role and the candidate market. Christian & Timbers helps companies establish the mandate and reach candidates whose experience matches the work, including elite FDEs who remain largely invisible to conventional searches.