The Forward Deployed Engineer: a guide for career advisors and CS departments
In brief
Aaron Levie said career counselors should quickly figure out how to get students into forward deployed engineer roles. The role exists, it's exploding, and it pays $150K–$700K+ total comp. The career infrastructure just hasn't caught up yet. Here's what to tell students.
Contents
In May 2026, Aaron Levie — CEO of Box and one of the more consistently accurate observers of enterprise technology — posted that career counselors "need to quickly figure out how to help students get forward deployed engineer jobs."
He's right. The role is real, the demand is massive, and the career infrastructure — career centers, CS departments, bootcamp curricula — hasn't caught up. Students are graduating into an industry that is desperately hiring for a role most advisors have never heard of.
This guide is what career advisors should know, and what to tell students.
What an FDE actually is
In plain terms: a Forward Deployed Engineer is an engineer who moves inside a client company for weeks or months, builds production AI systems against their actual technical environment, and leaves something that works.
The key word is "production." This is not consulting. A consultant leaves a document recommending what to build. An FDE leaves a deployed, running system — connected to the client's real databases, their real APIs, their real infrastructure.
The "forward deployed" framing is military in origin — it means stationed in the field. For engineers, it means operating inside the client's environment rather than building from the outside and handing over software. The engineer is physically and organizationally inside the problem.
A useful comparison table for explaining this to students:
| Role | What they leave behind | Build code? | Client environment? |
|---|---|---|---|
| Strategy consultant | A recommendation | No | Rarely |
| Solutions engineer | A demo | Sometimes | Occasionally |
| Implementation partner | A configured product | Sometimes | Yes, but using vendor tools |
| Forward Deployed Engineer | Production AI systems | Yes | Yes, against their actual systems |
The simplest way to explain it: imagine a senior engineer who spends six months inside your company, gets access to all your systems, and builds the AI infrastructure you couldn't build yourself. That's the FDE model.
Why this role is exploding right now
The demand surge has a specific cause. AI models are now genuinely capable. The problem is deployment.
When a 50,000-person pharmaceutical company wants AI to help their clinical trial managers, the challenge isn't finding a good AI model. The challenge is that the relevant data lives in three different systems that don't talk to each other, the APIs are undocumented, the compliance requirements restrict how data can flow, and the people who will use the system aren't engineers. No AI company can solve that by shipping better software. Someone has to go in.
Every major AI company has now concluded the same thing:
- Anthropic launched a $1.5B joint venture with Blackstone, Goldman Sachs, and Hellman & Friedman to deploy Claude via embedded engineers
- OpenAI launched "DeployCo" — a $10B operation — and immediately acquired Tomoro, a 150-person FDE firm, to staff it
- EY launched FDE roles in the UK in April 2026, expanding globally
- Accenture launched an FDE practice with Microsoft and ServiceNow
- Deloitte, Capgemini, and Google have similar programs in motion
FDE hiring is up 800% since January 2025. fwddeploy.com launched as a dedicated job board for this role category. When a role has its own job board, it has become a category.
This is not a niche. This is the emerging labor market for technical AI deployment.
Who should pursue this
The student profile that maps well to FDE work:
Applied technical background — CS major or minor with a strong focus on building things, not just theory. The signal is internships where they shipped something real that users touched, not just coursework.
Shipped something real — The best FDE candidates have delivered production code in conditions with real constraints. A summer job where they built a small internal tool that the company actually uses. A freelance project for a local business. Anything where there was a real stakeholder and a real deployment.
Can talk to non-engineers without condescension — This is rarer than it sounds. Many technically strong students have never practiced explaining their work to someone who doesn't code. Career advisors can help here: does the student have any experience teaching, tutoring, or working in cross-functional teams?
Self-directed — FDE work happens in conditions where full requirements are never given. Students who need structured direction before they start will find this role extremely uncomfortable. Students who thrive on ambiguity and have a history of figuring things out independently are natural FDE candidates.
Interested in how businesses operate — The best FDEs are genuinely curious about how companies work, not just how code works. Students who have done business-facing internships, read about company strategy, or shown interest in how technology decisions get made organizationally are good candidates.
Who is probably not the right fit: Students who want a quiet, internally-focused engineering role. Students who find client interaction draining. Students who need full specifications before they start work.
What to encourage them to build
Specific curriculum guidance for career advisors:
Agent development — not just prompting
The distinction matters. Prompting is writing better text to get better AI outputs. Agent development is building systems where AI takes actions, uses tools, and operates over multiple steps. The second is what FDEs actually build. Courses or workshops that cover the Anthropic or OpenAI APIs at the implementation level — writing code, not using the chat interface — are the relevant ones.
MCP servers and data integrations
Model Context Protocol is how Claude connects to external data sources. Every FDE at Anthropic is building these. Students who have built an MCP server connecting real APIs demonstrate a specific, current, in-demand skill. This is buildable in 2–3 weeks of focused work with the right resources. The MCP server overview article covers the concepts.
Production deployment basics
Authentication, error handling, observability, basic monitoring. The difference between "runs on my laptop" and "runs in production when someone uses it at 2am." This is teachable through project-based work where students actually deploy something — even small — to a real environment.
Business communication practice
One specific exercise that is highly effective: have students write a technical explanation of something they built for someone who doesn't code. One page. The exercise surfaces the gap between "can code well" and "can communicate about technical work to non-technical audiences." FDEs need both.
Where to find these jobs
For students actively looking:
Directly at AI companies:
- Anthropic Forward Deployed Engineer, Applied AI — the gold standard job description for this role; worth reading as a benchmark even if students don't apply immediately
- Palantir Forward Deployed Software Engineer — the original role, hiring from universities
- OpenAI (via DeployCo — new roles being added frequently)
- Scale AI FDE team
At consulting and professional services firms:
- EY (active expansion, entry-level roles available)
- Accenture (Microsoft and ServiceNow FDE practices)
- Deloitte AI Center of Excellence
At enterprises directly:
Large banks, pharmaceutical companies, logistics firms, and government contractors are building internal FDE functions. These roles often appear under titles like "Applied AI Engineer," "Enterprise AI Engineer," or "AI Implementation Engineer." Worth encouraging students to search these phrases specifically.
Job boards:
- fwddeploy.com is the most comprehensive FDE-specific job board
Compensation expectations to set:
- Entry-level FDE at major AI companies or Big 4: $150K–$250K total comp
- Mid-level: $250K–$450K
- Senior: $450K–$700K+
These are not inflated estimates. They reflect genuine market scarcity for the combination of technical depth, client capability, and AI expertise the role requires.
How to evaluate training programs
Several training options have emerged as the FDE role has grown:
- FDE Academy — purpose-built for FDE preparation
- IIT Roorkee and Supervity have launched FDE-specific programs
- Various bootcamps are adding "AI deployment" tracks
What to look for in any program:
- Hands-on deployment practice — students should build and deploy real systems, not just watch lectures
- Real enterprise case studies — exposure to messy real-world environments, not just clean examples
- Portfolio outcomes — the program should produce work students can actually show to hiring managers
What to be skeptical of:
- Pure theory without implementation
- No client simulation component
- A certificate without a portfolio of work
- Programs that end at "prompting" without covering agent development and deployment
The credential itself matters less than what the student can demonstrate. An FDE candidate with a strong GitHub portfolio and no certificate is more competitive than one with a certificate and nothing to show.
One thing to tell students right now
Apply to Anthropic's Forward Deployed Engineer, Applied AI role directly.
Read the job description in full. It is the clearest public specification of what this role requires — what skills, what deliverables, what experience. Even students who aren't immediately ready to apply should read it as a roadmap.
The role is public. It's hiring. And for students who have built the right technical stack and have a portfolio that shows they can operate in messy real-world environments, it's attainable within 12–24 months of focused preparation.
The career infrastructure for this role is still being built. Career advisors who understand FDE hiring now will be the ones students trust with their career decisions over the next several years. The demand is not a trend — it's structural. Enterprises need engineers who can deploy AI. That need will compound as AI capabilities continue to improve.
Aaron Levie was right: career counselors should be figuring this out. This guide is a start.
Further reading for students
- What is a Forward Deployed Engineer — the full definition, what FDEs build, compensation ranges, the hiring landscape
- How to become a Forward Deployed Engineer — the four-part skill stack, the three transition paths, and what the interview actually looks like
- The 5 portfolio projects that signal FDE readiness — specific, buildable projects that demonstrate FDE capabilities to hiring managers