Become the engineer who deploys the AI.
Twelve weeks of hands-on skills building, fieldwork with real businesses, live AI deployments, and interview practice — with experienced Forward Deployed Engineers.
Apply for the next cohortSee the pathThe job every AI company is hiring for — and can't fill.
A Forward Deployed Engineer takes an AI system out of the lab and into a customer's actual operation: finds the problem, scopes it, builds against real data and real channels, proves it works, and reports the result in business terms.
Most people close to this work already have some of the skills, but not all of them. Engineers have the technical depth and have never sat across from the customer. Solutions and sales engineers know the customer and have never owned the build. Almost nobody has been given the chance to apply what they have in a situation that is actually AI-native. The program closes the gaps you have, and gives you that hands-on experience.
growth in Forward Deployed Engineer job postings, January to September 2025.
committed by the largest tech and AI companies to dedicated AI deployment units in 2026 alone.
typical total compensation for FDEs at AI companies; far higher at the frontier labs.
Based on learnings from.
The program draws on first-hand experience with how these companies build, deploy, and hire for Forward Deployed work. They are not partners or sponsors of the program.
Coinbase
DeployCo
Fireworks AI
Decagon
LiveKit
PolyAI
Bland
Casca
Poetic
Bretton AI
AfterQuery
Alleviate Health
Trustwise
Luzid
What you'll learn.
The concepts the program teaches, across both tracks. The material itself lives inside the program.
Agent architecture
Tool use, memory, orchestration, and where a model belongs in a system versus where it doesn't.
Voice and real-time channels
Telephony, latency budgets, and the engineering of systems that have to answer in under a second.
Evaluation frameworks
How AI systems are tested before and after deployment, and how to prove one works to someone who doesn't trust it.
Determinism vs. inference
Making what can be deterministic deterministic, and reserving the model for the ambiguous parts.
Integration and deployment
Connecting to the tools a business actually runs — phones, forms, CRMs, booking systems — and shipping into production.
Human-in-the-loop design
Exception paths, escalation, and the fallbacks that make an automated system safe to leave running.
Discovery
Running a conversation with a non-technical stakeholder that surfaces the real problem and the number attached to it.
Scoping and decomposition
Breaking an ambiguous business problem into a buildable path, with explicit cuts and explicit tradeoffs.
Design documents
The one-page architecture and decision record written before anything is built.
Measuring impact
Choosing the metric, capturing the baseline, and attributing the change honestly.
Case-study writing
Turning a deployment into a document a hiring manager or executive reads in three minutes and believes.
The FDE interview loop
Decomposition rounds, take-homes, work simulations, and how to position real work in each of them.
What you leave with.
A certificate of completion, and everything behind it. Twelve weeks of real work produces a body of evidence — the kind a hiring manager can open, read, and verify.
A live deployment
An AI system running inside a real business, on their phone line, inbox, or workflow — designed, built, integrated, and shipped.
A portfolio of builds
The hard-skills phase ships working systems — agents over real channels, an evaluation harness, an automated workflow.
Discovery field notes
Your conversations with real business owners, written up: what you asked, what you heard, what it costs them, and why one problem made the cut.
A scoped brief and design doc
The problem statement, the number attached to it, what's in and out of scope, and the one-page architecture.
An evaluation suite and results
The test suite that proves your system works, the baseline you measured against, and the numbers after it went live.
A written case study
Problem, number, scope, architecture, evaluation, result — in the house style FDE teams write themselves.
Interview tape and feedback
Every simulation recorded — decomposition rounds, design reviews, full loops — with written feedback after each, from your first week to your last.
A rebuilt profile and pitch
Resume and positioning rewritten around the deployment, plus the internal pitch or external narrative, ready for whichever move you're making.
A certificate of completion
Issued when all program deliverables are done, with the syllabus and hours behind it.
The path. Four phases, two tracks.
The build track moves through four phases of three weeks each. The interview track runs alongside it the whole way, light at the start and intensive by the end.
Hard skills development
The technical foundations of deployment work: agents that operate over real channels, evaluation frameworks, and the judgment to decide what a model should and shouldn't do.
Soft skills development
Discovery as a discipline. How to sit with a non-technical stakeholder, surface the problem that actually costs them something, and scope it into something buildable.
Full deployments
Design, build, and deploy a complete AI system for a real organization, integrated with the tools they already run, with evaluation and a human fallback behind it.
Documentation and communication
Measure the outcome, write it up the way deployment teams do, and learn to present technical work in business terms — to interviewers, managers, and customers.
How it runs.
One-to-one
No course library. Experienced Forward Deployed Engineers work the plan with you in regular live sessions, and review everything you ship.
Recorded and graded
Every interview simulation is recorded and returned with written feedback. Every build has a rubric. You can see the slope.
Built for people with jobs
Six to eight hours a week, scheduled around your work. Sessions are set week by week, and the program flexes when your calendar does.
How admission works.
The program is one-to-one, so seats are limited by the instructor's calendar and filled by fit, not by order of application.
Apply
Tell us about your background, what you've built, and what you want out of the next twelve weeks. Takes five minutes; you'll hear back within a day.
Fit call
A thirty-minute conversation about your background, your target, and whether the program is the right lever for it. Some people are better served elsewhere; we'll say so.
Plan and start date
If it's a fit, you receive a personalized twelve-week plan built around your background and target, and you choose a start date.
Employer-sponsored participants.
Many participants have the program covered through a learning-and-development or tuition-reimbursement budget. Everything an approver typically asks for is provided.
- Program syllabusPhases, learning outcomes, and the artifacts produced, in a form suitable for submission.
- Itemized invoice and receiptsIssued by DialTech Ventures LLC, in the participant's or the company's name.
- Stated time commitmentApproximately ninety instructor-guided hours across twelve weeks.
- Certificate of completionIssued on completion of all four program deliverables.
- Manager summaryA one-page explanation of what the company gets, written for the person approving it.
Every engineer will be a deployment engineer. Most haven't been taught how.
The value of AI is realized at the point where it meets a real operation, and the people who can carry it there are the scarcest talent in the industry. FDE Dispatch exists to teach that work the only way it can be taught: by doing it, for a real business, with someone who does it for a living.
Apply for the next cohort.
Applications are reviewed as they arrive. If it looks like a fit, you'll hear back within a day with times for a short call, and we build your twelve-week plan from there.
The Dispatch.
Field notes from live AI deployments, frameworks for learning the work, deployment case studies, what FDE interview loops are actually testing, and who's hiring.