FDE Handbook
What is a Forward Deployed Engineer?
The definitive guide to the role enterprises need to move AI from pilot to production, and almost nobody trains for.
In short: A Forward Deployed Engineer (FDE) is an engineer who embeds inside a customer's organization, learns the business domain, and ships governed AI and agentic systems in production while transferring capability to the client's team.
Last updated July 2026
What does an FDE do day to day?
An FDE is not a demo engineer and not a staff aug contractor with a vague ticket queue. They sit inside the client's environment: on-site or in-VPC, and own the path from problem discovery to governed go-live.
Typical work includes domain interviews, context engineering across source systems, MCP and API integration, agent workflow design, security review with stakeholders, production deployment, and pairing with client engineers so knowledge stays when the engagement ends.
- Discover the highest-value use case inside the business, not the flashiest demo
- Build and govern context layers agents can trust
- Configure multi-agent workflows with appropriate models and tools
- Ship to production inside client security boundaries
- Document, hand off, and train client teams
Where did the FDE role come from?
The Forward Deployed Engineer title was popularized by Palantir, where engineers embedded with government and enterprise customers to deliver mission-critical software in complex environments. The pattern: embed, learn, ship, transfer: proved durable because it solved a problem generic consulting could not: production ownership inside the customer boundary.
Today every AI-forward company faces the same structural gap. Models are commoditized; the bottleneck is engineers who can embed, integrate, and operate agentic systems where data, policy, and politics meet.
What skills does an FDE need?
FDEs are strong software engineers first. They also need client-facing judgment, systems thinking across messy enterprise stacks, and comfort operating with incomplete information.
- Software engineering: Python/TypeScript, APIs, cloud, CI/CD
- Context engineering: retrieval, MCP, knowledge integration
- Agent design: orchestration, tool use, guardrails
- Enterprise fluency: security, compliance conversations, stakeholder management
- Agency: finding and driving work without constant direction
How do you become an FDE?
Most FDEs today come from strong engineering backgrounds and learn embedding on the job, often through trial and error over years. FDE Factory compresses that path into Cohort 02, 12 weeks with partner-graded production work.
Cohort 02 accepts 40 engineers. The path is application, readiness assessment, panel interview, 12-week program, certification, and first deployment.
A day in the life of an FDE
No two days look the same. That's the point. Mornings might be a working session with the client's ops team mapping a workflow. Midday could be pairing on an MCP connector to their document store. Afternoons often land in security review or a demo to a skeptical stakeholder who needs to see production constraints, not a slide.
FDEs spend less time in standups about tickets and more time reducing uncertainty: Which use case actually matters? What context does the agent need? What would block go-live next month? The job is part engineer, part product thinker, part diplomat.
- Stakeholder interviews and workflow mapping
- Context layer design: retrieval, MCP, permissions
- Agent configuration and integration testing
- Production deployment and runbook writing
- Pairing with client engineers for capability transfer
FDE career path
FDEs often enter from senior software engineering, platform, or solutions-adjacent roles. The trajectory runs from embed engineer to lead FDE on larger programs, then to pod lead or internal platform roles at AI-forward companies.
Compensation tends to reflect embed intensity, domain complexity, and production scope. See our salary guide for factors, not guarantees.
Who hires FDEs?
AI platform vendors, enterprises running internal agent programs, and consultancies building embed practices all hire FDEs. The common thread is production accountability inside the customer boundary, not demo-only delivery.
Companies that hire FDE Factory pods get Cohort 02-trained engineers with shared methods and FDEE coverage. See our employer guide for the engagement model.
Frequently asked questions
Is an FDE the same as a field engineer?
Field engineers often focus on hardware, infrastructure, or support at customer sites. FDEs specifically build and deploy software, especially AI and agentic systems, in the customer's business context.
Do FDEs write code in production?
Yes. FDEs are hands-on builders. They commit code, configure agents, integrate systems, and own go-live, not just architecture documents.
Can remote engineers be FDEs?
Embedding can be on-site or inside a client VPC with strong communication discipline. The defining trait is ownership inside the customer's environment, not necessarily physical presence every day.
How is an FDE different from a solutions engineer?
Solutions engineers support pre-sales with demos and POCs. FDEs own post-sale production delivery: domain discovery, context layers, go-live, and capability transfer inside the client environment.
Do I need an ML PhD to be an FDE?
No. Strong software engineering, systems thinking, and client-facing judgment matter more than research credentials. You need to ship governed systems in messy enterprise stacks, not publish papers.
What stack do FDEs use?
Cohort 02 trains on the agentic stack: MCP for integrations, context engineering patterns, multi-agent orchestration, and production eval tooling, always inside partner environments with real constraints.
How do I train to become an FDE?
Apply to Cohort 02 at FDE Factory, 40 seats, 12 weeks, partner-graded capstone. See the Cohort 02 page and training program for the full syllabus.
What is an FDE pod?
An FDE pod is a small team of FDEs and FDEEs deployed together: builders and eval engineers trained on the same methods. Companies hire pods when they need production outcomes faster than traditional search or consulting.