Engineers trained to embed.
Ready in days.
We train Forward Deployed Engineers and Eval Engineers on the agentic stack: then place them where most AI pilots stall: the long gap between demo and production.
[ 01 / 05 ]
AI doesn't fail in demos. It fails after launch.
Most companies don't lack models. They lack engineers who can sit inside the business, learn the domain, and carry AI from pilot to governed production. Consultancies often leave at go-live. What companies need is a person who stays, and knows how to ship. That's who we train.
[ 02 / 05 ]
Two paths. One standard.
Forward Deployed Engineer
You build in production: learning the client's domain, shipping governed agentic systems in their environment, and helping their team take ownership as you go.
- · Context engineering & MCP integration
- · Agent configuration & orchestration
- · Enterprise VPC deployment
- · Client discovery & capability transfer
Forward Deployed Eval Engineer
You keep production agents accurate, safe, and compliant after go-live: evaluation, guardrails, and continuous testing. A discipline almost nobody else teaches.
- · Eval harnesses & continuous testing
- · Guardrails & policy controls
- · Drift detection & monitoring
- · Compliance reporting
[ 03 / 05 ]
Twelve weeks, start to finish.
Every project is judged on one question: would this survive inside a real organization?
Embed & discover
You learn the client's world: interviews, domain research, and finding the use case that actually matters. We treat this as a skill you practice, not a slide you present.
Unify context
You connect source systems into one governed context layer: MCP integration, knowledge graphs, and the plumbing production agents need.
Configure agents
You build and tune multi-agent workflows inside a live partner environment, with evals, guardrails, and the right model mix for the job.
Deploy & monitor
You ship to a real environment, then keep improving with monitoring and evals. Your final project is graded by the partner you're working with, not by us alone.
[ 04 / 05 ]
Hire a pod, not a pile of resumes.
Instead of searching the open market for a role that barely exists, you get a trained FDE + FDEE pod embedded with your team in days, already fluent in the agentic stack and how we deliver in production.
- ✕Six-month searches for a role with almost no talent pool
- ✕Consultants who excel at slides and leave at go-live
- ✕No one watching quality once the system is live
- ✕Knowledge that walks out when the contract ends
- ✓Trained pods embedded in days, not quarters
- ✓Engineers who build in production and stay accountable
- ✓FDEEs who keep agents accurate and compliant after go-live
- ✓Your team learns alongside the pod, not after they leave
[ 05 / 05 ]
Ready to become an FDE?
We take 40 people in Cohort 02 because learning works best in a strong room. You arrive as a capable engineer. You leave certified, with real deployment experience, and a profile companies are actively looking for.
look for
We can teach the stack. These, we look for.
An FDE works alone inside someone else's company, on problems nobody has fully scoped. We screen for the three traits that matter most in that environment.
Agency
At a client site, nobody hands you a ticket queue. You find what matters, decide, and move. If you need permission for every step, the work stalls, and so do you.
Risk-taking
Production means committing before you're certain: demoing to a skeptical exec, choosing an approach with incomplete information. The job is calculated bets, not endless hedging.
Resilience
Deployments break. Stakeholders push back. Week nine sometimes means a rebuild. The people who get AI past the pilot graveyard are the ones who take the hit and keep shipping.
Stories from deployments.
Representative example, details anonymizedClaims triage agents, global insurer
A pilot had stalled for seven months. An embedded FDE + FDEE pod took it to governed production in nine weeks: 34% faster claim cycles, with evals running continuously since go-live.
“The pod understood our stack, compliance constraints, and data model by week two. We've since hired two of them full-time.”
“Twelve weeks here did what would have taken me years of trial and error on the job.”
compliance
Built for teams who ask hard questions.
Audit in progress. We can share the report under NDA when it's complete.
Pods work inside your perimeter: your cloud, your keys, your controls.
We never use client data for training. It's purged when the engagement ends.
All work product belongs to you. We sign a standard MNDA before any embed.
New to forward deployed engineering? Start with the handbook, Cohort 02, or what is an FDE.
What is a Forward Deployed Engineer (FDE)?
A Forward Deployed Engineer (FDE) is an engineer who embeds inside a customer's organization to learn the business domain, ship AI and agentic systems in production, and transfer capability to the client's team. FDEs work on context engineering, agent configuration, integration, and go-live, not slideware or isolated demos.
What is a Forward Deployed Eval Engineer (FDEE)?
A Forward Deployed Eval Engineer (FDEE) keeps production AI agents accurate, safe, and compliant after deployment. FDEEs build eval harnesses, guardrails, drift detection, and continuous testing: the discipline required once a system is live.
What's an FDE pod?
An FDE pod is a small team of certified FDEs and FDEEs deployed together inside a client environment. Pods combine build and evaluation capability in one embed, trained on the same stack and delivery method.
What is Cohort 02?
Cohort 02 is FDE Factory's current open intake, 40 seats across FDE and FDEE tracks over 12 weeks of partner-graded production work. It is the only cohort accepting applications today.
How long is FDE Factory training?
Cohort 02 runs 12 weeks covering discovery, context engineering, agent build, and production deployment. Graduates earn FDE or FDEE certification depending on track.
How is FDE Factory different from a bootcamp?
Bootcamps teach general software skills. FDE Factory trains a specific operational role: embedding in enterprises, shipping governed agentic systems, and working alongside FDEEs on evaluation and compliance. Cohort 02 is capped at 40 seats with partner-graded final projects.
How is an FDE different from a solutions engineer?
Solutions engineers focus on pre-sales demos, POCs, and technical win support. FDEs own production delivery inside the client environment after the deal: domain discovery, context layers, agent orchestration, go-live, and capability transfer.
Do companies depend on FDE Factory forever?
No. Capability transfer is built into the pod model. Your team gains operational knowledge as the pod ships. The goal is independence, not permanent vendor reliance.
How fast can a pod embed?
Trained FDE + FDEE pods can embed in days, not quarters, because they arrive with shared methods, stack fluency, and hands-on experience from Cohort 02 partner environments.
Is Cohort 02 remote or in-person?
Cohort 02 training is structured for remote delivery with live partner stack work. Deployments after certification embed on-site or inside client VPCs depending on engagement requirements.
How do I apply to Cohort 02?
Submit the application form at fdefactory.ai/apply. You'll receive an async readiness test within five business days, followed by a panel interview for selected candidates.
How do I hire Forward Deployed Engineers?
Request a trained pod through our company inquiry form, or ask us to match Cohort 02 graduates to your stack after capstone completion.
What does FDE certification mean?
Certification confirms completion of Cohort 02 and passing a partner-graded capstone: demonstrating you can ship governed agentic systems inside a real organization, not just pass a multiple-choice exam.
Can I switch between FDE and FDEE tracks?
Track preference is set during Cohort 02 admissions. Some engineers cross-train over time, but certification reflects the track completed in the cohort.
Where should I start learning about FDEs?
Read our handbook: start with What is an FDE, then the glossary, Cohort 02 program page, and insights essays on pilot-to-production and context engineering.