- A marker nobody documented
- Your brief
- The model, the data and the evidence
Build the AI Act technical file for an exam-marking model
Kestrel Learning marks 1.4 million short answers a year with a model nobody has documented. Classify it under the EU AI Act, test it for disparity across candidate groups, design the human oversight that makes it lawful, and assemble the Annex IV technical file — including the finding the product director will not want.
PCREATED BYPrepmeetsProject studio, Prepmeets
THE CHALLENGE → YOUR OUTCOME
Real context. Work that matters.
Kestrel’s marker has been in service for three years and has never been assessed. You classify it, reconstruct the documentation the Act requires, measure disparity across first language and support arrangements, design the oversight that makes automated marking defensible, and tell the board plainly what you found — including the part that is expensive.
What’s at stake
Your responsibility: Work as the ai governance lead—investigate the requirements, implement your approach and verify the result.The marker was built in 2023 by two data scientists, both now elsewhere, and it works well enough that the exams team trusts it. It scores short free-text answers against a rubric; scores the model is unsure about go to a human marker, and a 12% sample is moderated. It cut marking turnaround from six weeks to nine days, which is why nobody has questioned it.
Success means: A defensible classification with reasoning, an Annex IV technical file complete or honestly gapped, a disparity analysis whose uncertainty is stated, an oversight design that is specific enough to implement, and a board note that makes the decision rather than deferring it. Explain the decisions and checks behind your implementation.
WHAT YOU’LL BUILD
01EU AI Act
02Model cards
03Fairlearn
Kestrel Learning automated short-answer marker
INCLUDED WITH YOUR PROJECT
A workspace, not just a document.
Your brief, a Jira-style task board and an interview studio—all in one place.
A look inside your workspace
Start with the full project brief
Build step by step Track your progress Own your interview story
Know where you’re heading.
- 01A marker nobody documented
- 02Your brief
- 03The model, the data and the evidence
- 04How the artefacts fit together
- 05Five phases
- 06Reviewing your own file
- 07What you hand over
- 08Submission
Practise explaining your decisions
Which decision mattered most in your solution, and how did you verify it?
Use your own implementation to connect Situation, Task, Action and Result into one interview story.
Before you begin · tools & running costs
Prerequisites: Your own accounts and access to Fairlearn, EU AI Act Annex IV structure, Model card, Human oversight design.
Additional costs: Cloud and tool usage in your own accounts may incur charges. Set a budget and clean up after testing.
Work in your own accounts and repositories. Never share credentials or confidential data.
What’s included
Project brief, target-design guidance, assigned tasks, resource guidance and interview prompts. Detailed requirements become part of your purchased workspace.
P
MEET YOUR PROJECT EXPERT
Prepmeets
Project studio, PrepmeetsWe are the team that builds Prepmeets. Between us we have run cloud platforms, delivery pipelines, analytics teams and AI governance programmes in banks, hospitals, retailers and government departments, and these projects are the work we used to hand to new joiners in their first fortnight. Every one of them is a real scenario with the client’s name changed, scoped to a week of evenings, and reviewed by the person who originally did the job.
View verified expert profile Expert review is an optional paid service, separate from project access.
£19.90Start this project