Prove what you can build with AI.
Proving Ground is a free, open library of work challenges for the eight roles AI has changed most. Each challenge is scored against a published rubric, in bands, with a video walkthrough that becomes a public artifact you own.
Name the Failure Mode
Scored against the bar
Written feedback
on all five dimensions
Video walkthrough
a public artifact you own
Profile entry
added to your Provn Profile
How the Proving Ground works
Four steps. The last two are the parts nobody else does.
Pick a role and a tier
Eight role ladders, four tiers each, from a 20-minute Practice challenge to a 40-minute Hard tier. Every challenge in a ladder shares one scenario, so the four submissions read as one body of work.
Do the work and record your video
Real constraints, real deliverables, and a video walkthrough where you explain your decisions and the one moment you pushed back on the AI. Use whatever tools you use at work.
Get scored against the published bar
Every dimension and weight is on the page before you start. You get a band, specific written feedback per dimension, and a Profile entry.
Be seen by top employers
Your scores, feedback, and video walkthroughs build a portfolio that proves your AI skills. Score 80+ and you are a top candidate that employers are seeking.
Eight roles. Eight ladders.
View a role page to learn what the job has become, what the top employers post and pay, the skills that show up across their job descriptions, and the ladder that proves you can do the work.
AI Product Manager
~12,400 US AI-product postings; top ten employers post $171K to $460K
- Posted pay
- $171K to $460K
- Ladder
- 4 tiers, 20 to 40 min each
- Start with
- Name the Failure Mode (20 min)
AI / Agentic Software Engineer
#1 fastest-growing US job title; postings up 143% YoY
- Posted pay
- $147K to $625K
- Ladder
- 4 tiers
- Start with
- What Breaks in Production (20 min)
Forward-Deployed Engineer
Postings up 800 to 1,165% in 2025; two ladders, one role family
- Posted pay
- $135K to $330K
- Ladder
- 8 tiers (customer-facing + full-stack)
- Start with
- The Ask Behind the Ask (20 min)
Data Analyst / Data Scientist
The one employer set hiring at volume: 100+ open reqs at Amazon alone
- Posted pay
- $131K to $370K
- Ladder
- 4 tiers
- Start with
- The Metric That Lied (20 min)
Product Designer (AI-native)
AI-model design reqs live at frontier labs and design-tool companies
- Posted pay
- $130K to $460K
- Ladder
- 4 tiers
- Start with
- Design the Wrong Answer (20 min)
Growth / AI-Enabled Marketer
Net-new functions (GEO, agent-stack marketing) appearing on boards
- Posted pay
- $150K to $320K
- Ladder
- 4 tiers
- Start with
- Find the Broken Step (20 min)
Account Executive / Solutions Engineer
23 AE reqs live at one AI lab; proven Provn placement role
- Posted pay
- $171K to $435K OTE
- Ladder
- 4 tiers
- Start with
- The First Ninety Seconds (20 min)
Ladder opens October 2026
Program Manager / TPM / AI Ops
AI-launch TPM reqs at every frontier lab
- Posted pay
- $120K to $445K
- Ladder
- 4 tiers
- Start with
- The Step That Shouldn't Be Automated (20 min)
Ladder opens October 2026
Posted pay is the range on specific requisitions read from each employer's own careers board in August 2026. It is not a market median and not an offer. Each role hub lists the requisition, level, and read date.
The bar, published
Every Proving Ground rubric is built on the same spine, so a score on one ladder means the same thing as a score on another. The weights flex a little by tier; the dimensions do not.
Advance at 75. Draft Board at 80.
- Transformative90 to 100
- Adoptive80 to 89
- Capable70 to 79
- Below Capable60 to 69
- Not Yetunder 60
You see your band and your feedback. Employers see numeric rankings only inside a real hiring process. That split is deliberate: it keeps your development report separate from anyone's hiring decision.
| Dimension | What it measures | Weight |
|---|---|---|
| Core role execution | The domain deliverable itself: the memo, the code, the design, the analysis | 30 to 40% |
| AI-specific craft | The role's AI dimension: eval design for a PM, agentic development for an engineer, AI-research verification for a seller | 20 to 25% |
| Strategic judgment | Prioritization, scoping, and the honest handling of constraints | 15 to 20% |
| Video walkthrough & communication | Can a colleague act on your explanation alone? Structure and reasoning, never accent or polish | 20 to 25% |
| AI fluency | Directed, iterative, critically evaluated AI use, shown in your usage log and your video. Cannot compensate for weak core work | 10 to 15% |
The written deliverable can be generated. A five-minute explanation of your own choices, including where you overruled the AI, cannot. That is why the walkthrough is mandatory on every challenge, and why it carries up to a quarter of the score.
Start with a 20-minute challenge.
Every ladder opens with a Practice tier you can finish today. Read it first. Sign up when you are ready to submit.
FAQ
Is the Proving Ground free?
Yes. Every challenge is free to read and free to take. A free Provn account is needed to submit work and receive a score; it is never needed to read a challenge, a rubric, or a role page.
Do I need an account to read a challenge?
No. Every challenge, its constraints, its deliverables, and the dimensions it is scored on are public. You need an account only to submit and to save work to your Provn Profile.
How is my work scored?
Each challenge publishes its scoring dimensions and weights up front. Submissions are scored with AI against that rubric and reviewed by a human before results go out. You receive a band (Transformative, Adoptive, Capable, Below Capable, or Not Yet) plus specific written feedback per dimension.
Can I use AI to do the challenge?
Yes, and we expect you to. We score how you use it, not whether you use it. Every challenge asks for a short AI usage log and a mandatory moment in your video where you explain one place you pushed back on or redirected what the AI gave you.
Can employers see my score?
You see your band and your feedback. Employers see rankings only inside a real hiring process they are running on Provn, and they never see your Proving Ground work unless you apply to their role or opt your Profile into visibility.
What is the Draft Board?
The Draft Board is the group of builders whose challenge work scores 80 or above. Employers hiring on Provn start there. Every Proving Ground ladder feeds it.
Why is the video walkthrough mandatory?
Because it is the one artifact that cannot be faked. A written deliverable can be generated; a five-minute explanation of your own decisions, including where you overruled the AI, cannot. Hiring managers tell us the video is what makes them feel they already know a candidate before the first call.
What if my score is Not Yet?
You get the same specific feedback as everyone else, and you can retake the challenge in the next cohort. Start with the Practice tier of the ladder, which is built to be finished in 20 minutes.