FDE at Notion, Ex-Google, OReilly Author

You can build. That was never the problem.
The problem is the round nobody prepared you for. A vague enterprise problem, no spec, 45 minutes, and a stranger watching you think. You proposed an architecture in the first three minutes because that felt like competence. You never heard back.
FDE postings went from 643 to over 5,300 in a year. The role moved faster than the prep did. That round has the lowest pass rate in the loop, around 40%, and at OpenAI case studies and business judgment are half the evaluation. Almost nothing you studied touched it.
Most courses would hand you more curriculum. That isn't the gap. You cannot rehearse scoping against a blog post, or prove judgment on a resume.
So we do reps. A real customer, broken data, stakeholders who contradict each other, and an ask that's partly impossible. You scope it, build a thin slice, and defend it while I play the VP who wants the wrong thing.
Twice, on camera, scored against a published rubric.
I wrote the handbook on this role. Check it out in the free resources
Go from strong engineer who freezes in the case round to candidate who scopes, ships and defends a real deployment under pressure.
Diagnose what the customer actually needs from four stakeholders who contradict each other twice in the same Slack thread (
Interrogate the constraints that decide a deployment: data access, permissions, latency, cost, and security review
Cut a six-month ask into a four-week scope, then defend it in a one-page memo scored against a published rubric
Build grounded retrieval over 140 messy contracts with superseded amendments, duplicate entities, and inconsistent formats
Enforce permissions at the retrieval layer rather than in a prompt, so your system survives a real security review
Judge when the right answer to "build me an agent" is retrieval with no agency at all, and say so out loud
Construct a 30 to 50 case labeled eval set with a pass definition you can defend to someone who did not build it
Test the cases that decide trust: unanswerable questions, permission-denied queries, and superseded contract data
Report failure honestly in a one-page memo, with cost per query and a clear statement of what you would not ship
Present your build to a VP who wants something you should not build, and hold the line without losing the room
Justify trade-offs across latency, cost, coverage, and governance while someone pushes back in real time
Do it twice on camera, scored on the same rubric both times, so you can see exactly what changed
Distinguish what Palantir FDSE, OpenAI FDE, and Anthropic Applied AI Engineer each weight, since the loops are not the same
Palantir leans ontology and data engineering. The AI labs lean production LLM systems: RAG, evals, agents, fine-tuning
Calibrate comp by company and level using live data from the FDE job board I run
Compose a deployment case study a hiring manager can read in four minutes and actually learn something from
Articulate "Why FDE?" so it sounds like your reason, not the answer every other candidate gives
Leave with the case study, six pressure-tested stories, and access to the FDE job board and alumni network

Founding FDE @ Notion | O'Reilly Author | Ex-Google, Bloomberg
Engineers targeting FDE roles
You can build. Now prove the judgment half of the loop nobody trained you for, on a real deployment, scored.
Solution engineers moving into FDE
You have the customer instincts already. Get the technical half graded by someone who runs the deployment
FDEs in their first year
You got the job and scoping still feels improvised. Get reps on broken data before you do it in front of a customer.
That's the bar. You don't need AI experience, a vector database, or a shipped agent. The starter repo handles the plumbing and we cover the
You can be two years in or ten, career-switching or already an FDE. What matters is that you'd rather build something graded than watch some
Your scoping gets reviewed and discussed. The people who improve most are the ones who get it wrong in week one and fix it in week two.

Live sessions
Learn directly from Paden Gayle in a real-time, interactive format.
Recruiter network access
I run the FDE job board these roles get posted to and work with the recruiters filling them. Alumni see openings 48 hours early. If you opt in, your case study and score go to my recruiter network after the cohort.
Two scored reps, on camera
You run the case round twice, live, graded on the same anchored rubric both times, and you see your delta. Nobody else on this platform will score you individually on this round.
The FDE Scorecard
The anchored rubric interviewers actually use, published so you can prep against it instead of guessing what good looks like. Yours to keep.
A real broken data room
140 messy carrier contracts, superseded amendments, duplicate entities, a permissions file, and four stakeholders contradicting each other. One of their four asks is impossible from the data. Finding that is the test.
Your deployment case study
You leave with a scoping memo, a working repo, a 30 to 50 case eval set, and a failure memo. A hiring manager reads it in four minutes and learns more than a resume tells them.
Lifetime access and free retakes
Every recording and material, forever. Rejoin any future cohort live at no extra cost. Lock in today's price even if your schedule slips.
Office hours and alumni community
Two open sessions per cohort, for questions and job leads after the course ends.
Maven Guarantee
Your purchase is backed by the Maven Guarantee.
4 live sessions β’ 8 lessons β’ 2 projects
Oct
27
Optional: Session 1: Why the case round kills strong engineers
Oct
29
Optional: Session 2: Decomposition lab β scored rep #1
Nov
3
Optional: Session 3: Defending under fire
Nov
6
Optional: The Gauntlet - Scored Rep#2
Live sessions
3 hrs / week
Tuesdays and Thursdays. All recorded. Sessions 2 and 4 are where you get scored on camera, so plan to attend those live even if you catch the others later.
Tue, Oct 27
11:00 AMβ12:15 PM (UTC)
Thu, Oct 29
11:00 PMβ12:00 AM (UTC)
Tue, Nov 3
12:00 PMβ1:00 PM (UTC)
Fri, Nov 6
12:00 AMβ1:00 AM (UTC)
Your deployment build π¨
4 hrs / week
Scoping memo in week one. Thin slice, eval set, and failure memo in week two. Everything is graded, everything is due before the session that discusses it. The starter repo means you spend this time on judgment rather than setup.
Reading and prep π
1 hr / week
Company loop cards, the scorecard, and short written pieces on retrieval over messy corpora, permission enforcement, and evals. Read before the session it belongs to. Nothing is longer than fifteen minutes.

Maven for Teams
Reimbursement
Get your company to pay
Everything L&D needs: email template, receipts, and certificate of completion.
Get reimbursedTeam discount
Learn with your teammates
Save 20%+ when 2 or more teammates enroll in the same cohort.
Save 20%+ with a teamPrivate cohort
Run a cohort for your org
A dedicated cohort with a custom schedule and curriculum, tailored to your team.
Book a private cohort$1,200
USD