Read the FDE Role Guide
FDE School
Join Waitlist

The curriculum.

A six-week program built to make you a Forward Deployed Engineer.

We take you from novice to expert in AI tooling in six weeks. You leave knowing how to ship real AI systems and ready to do it inside any company.

Join the Waitlist What the role actually is

The curriculum.

Six weeks. Three phases.

You become the AI native operator your company can't replace. No prior engineering background required.

Phase 01

Foundations.

You read a customer's system the way they read it themselves, and put a real number on the AI opportunity inside it.

Weeks 1 to 2 · 2 modules · Twice-weekly live with guest faculty

  1. 01Customer presence

    Customer discovery and ROI

    Week 1Module 1

    By the end of week one you can walk into a customer's operation cold, run a 90-minute discovery call, and walk out with a workflow map their own ops lead would sign. You turn vague pain ("our reps are slow") into a problem you can ship against. You know when the answer is an agent, when it is automation, and when it is a human in the loop. And you put a number on it. A north-star metric a customer commits to. An ROI hypothesis a CFO will actually engage with.

    Practical challenge
    Run a recorded discovery call with an assigned customer (instructor or community FDE playing a hostile VP of Ops). Submit the workflow map and a one-page ROI hypothesis within 48 hours.
    Portfolio deliverable
    A real workflow map of a customer operation, pain points scored on impact versus effort, one prioritized AI opportunity with a hypothesized intervention, and an ROI hypothesis a CFO would engage with.
  2. 02Technical depth

    Integrations and data

    Week 2Module 2

    By the end of week two you can wire two real production systems together and have the integration survive the failure modes that take real ones down. Auth, pagination, retries, idempotency, queueing. You treat them as design choices, not surprises. You read API docs the way an FDE reads them: looking for the lie. And when a customer drops a dirty Postgres dump on your desk with "the numbers don't match," you diagnose it from logs and queries without needing the engineer who built it. You become the person who can tell when the right move is fix the data, not fix the model.

    Practical challenge
    Integrate a mock CRM (HubSpot sandbox) with a Postgres-backed FastAPI service. Failures get injected mid-run. You handle them. A dirty Postgres dump lands the same week with a vague complaint ("the numbers don't match"). Diagnose, query, explain.
    Portfolio deliverable
    Bidirectional CRM to internal system integration, resilient to network failures and duplicate events, with a written failure-mode analysis and a diagnosis report on the dirty dataset.

Phase 02

The Stack.

You ship the AI system that doesn't crash in front of the customer.

Weeks 3 to 4 · 2 modules · Twice-weekly live with guest faculty

  1. 03Technical depth

    RAG and agents

    Week 3Module 3

    By the end of week three you can stand up a retrieval pipeline a legal team would actually trust, and an agent whose tool surface fails safely instead of catastrophically. You pick between function calling, MCP, and LangGraph by the constraint in front of you, not the marketing post in front of you. When the system hallucinates, you can tell whether it broke at retrieval or generation, and fix the right layer.

    Practical challenge
    Build an agent that answers questions from a packet of real, anonymized commercial contracts with mandatory inline citations.
    Portfolio deliverable
    An agent that queries internal docs with auditable citations, a written retrieval evaluation, and a chosen architecture justification.
  2. 04Technical depth and systems

    Production reliability

    Week 4Module 4

    By the end of week four you can take an AI system from "it works on my laptop" to running in a customer's cloud, with evals their procurement team will accept as an acceptance contract. You red-team for the failure modes that actually matter in production. You instrument the system so the next on-call engineer can find the failure without paging you. You plan the rollback before the first deploy, not after the first incident. And you sit in a security review without breaking into a sweat. This is the week that separates an FDE from a prototype-builder.

    Practical challenge
    Take the Module 3 agent. Write 50 golden cases. Ship a regression dashboard that fails the build on regression. Deploy the agent to a real cloud environment with proper observability, then run a planned rollback live in front of the cohort.
    Portfolio deliverable
    An evals suite with golden cases, a regression dashboard, a one-page eval contract that could be pasted into a customer SOW, a reproducible deploy with documented rollback plan, and a working observability dashboard.

Phase 03

The Operator.

You become the person the customer keeps in the room.

Weeks 5 to 6 · 1 module and the capstone build sprint · Twice-weekly live with guest faculty

  1. 05Customer presence

    Executive communication

    Week 5Module 5

    By the end of week five you can walk into a steering committee with a CFO, a CTO, and a VP of Ops and walk out with the next phase funded. You write design docs that unblock AI projects instead of stalling them. You translate technical trade-offs into the language an executive funds against, without dumbing them down. You know which decisions belong in the room and which ones you make yourself. The week you become the engineer customers ask for by name on the next project.

    Practical challenge
    Present the deploy from Module 4 to a panel playing a customer steering committee (CFO, CTO, VP of Ops). The panel grades on whether they would fund the next phase.
    Portfolio deliverable
    Technical design doc plus a 10-minute executive presentation, both delivered live and recorded.
  2. 06The capstone

    Capstone build sprint

    Week 6Capstone

    Six days of build. Demo day on the seventh. You pull every pillar together. Discovery, integrations, dirty data, agent, evals, deploy, executive narrative, business case. You ship a real system to a real cloud environment. You defend it live to a panel of working FDEs from the labs hiring this role. You walk off demo day with the artifact that lets you stop applying to FDE roles and start interviewing for them.

    Practical challenge
    Pick a domain (payments operations, B2B support, or revenue operations), ship a running system to a real cloud environment, and present a 15-minute demo with live Q&A to a panel of working FDEs.
    Portfolio deliverable
    Running capstone in production, architecture and design doc, evals suite with a passing dashboard, rollback plan executed at least once, business case with baseline and target metrics, and a recorded demo-day presentation.
Duration 6 weeks Five build weeks plus a capstone build sprint with demo day on the final day.
Format Twice weekly live Live with working FDEs as guest faculty, plus async build. Roughly 10 to 14 hours per week.
Cadence 4 cohorts per year Selective. We review every applicant. Apply early for the cohort you want.

Week 6. The capstone build sprint.

Your capstone becomes the artifact that gets you hired.

Seventy percent of AI pilots never reach production. Your capstone is the proof you ship the other thirty. It is the link you send a hiring manager instead of a resume. It is the system you walk an FDE interviewer through, live, with evidence. Not a story about a project you almost did.

The build forces every pillar to fire at once: discovery, integrations, dirty data, agent, evals, deploy, executive narrative, business case. None of it can be faked. Pick payments operations, B2B support, or revenue operations. The shape changes; the bar does not. What you walk out with is portfolio, recruiting collateral, and the first artifact a working FDE will look at when they decide whether to forward your name.

What you ship

  • Running system in a real cloud environment, accessible to instructors and demo-day attendees.
  • Architecture and design doc in the Module 5 format.
  • Evals suite with golden cases and a passing dashboard.
  • Rollback plan executed at least once during build.
  • Business case with baseline and target metrics.
  • 15-minute demo-day presentation with live Q&A.
  • Two-page executive summary written for a fictional customer CFO.

How it is graded

  • Does the system actually run end to end without instructor intervention? Pass/fail gate.
  • Are the evals real and is the regression dashboard honest? Pass/fail gate.
  • Does the design doc let a new engineer pick up the system cold? Rubric, 1 to 5.
  • Does the business case math hold up to a skeptical CFO? Rubric, 1 to 5.
  • Can you defend trade-offs under live Q&A from working FDEs? Rubric, 1 to 5.

Demo day is live, public, and recorded. Three working FDEs from partner companies sit on the panel. The recording becomes your recruiting collateral and the first look employers get when your name moves through the talent pipeline.

Cohort 01 opens Q3 2026

Six weeks from a seat in this cohort, you are a Forward Deployed Engineer.

Cohorts are small and we review every applicant. The waitlist is how you get the first look when seats open and the chance to start before the people still deciding.

Join the Waitlist