Portfolio artifact kit
Build the proof behind your AI role-fit decisions
Turn the first Academy modules into a recruiter-readable repository where every prompt, evidence rule, output, and limitation is open for review. Start from the public structure, then own the final decisions and limitations.
- Free public template
- No API key required
- Fictional examples
- README.md
- prompts/role-fit-assistant.md
- schemas/role-fit-data-card.schema.json
- examples/expected-response.md
{
"decision": "review",
"evidence": 4,
"open_questions": 2,
"confidence": "bounded"
}Inside the repository
Everything a reviewer needs to inspect your decisions
Each file answers a question a technical reviewer is likely to ask: what did you build, what evidence did it use, where can it fail, and how did you evaluate it?
- Included
README.md
Project brief, decision boundary, inputs, and failure modes.
- Included
prompts/role-fit-assistant.md
Evidence-led instructions for a bounded role-fit analysis.
- Included
schemas/role-fit-data-card.schema.json
Structured contract for requirements, evidence, and uncertainty.
- Included
examples/nordwind-job.md
Fictional role input used throughout the Academy exercises.
- Included
examples/amina-profile.md
Fictional candidate evidence with explicit known limitations.
- Included
examples/nordwind-data-card.json
Reviewed structured input matching the public JSON Schema.
- Included
examples/expected-response.md
Reference response showing evidence and uncertainty in practice.
- Included
evals/review-rubric.md
Evaluation criteria for evidence quality and product boundaries.
- Included
PROJECT_CHECKLIST.md
Final checks before publishing the repository as portfolio proof.
Four-step workflow
Move from example to your own evidence
Follow a short path from the public starter to a documented artifact that another person can inspect and discuss.
- 01
Fork
Create your repository from the public GitHub template.
- 02
Complete
Work through modules 1 and 2 with the fictional example.
- 03
Personalize
Replace the example with permitted role and candidate inputs.
- 04
Publish
Review limitations, document decisions, and share the result.
Built with an explicit boundary
The starter analyzes one manually provided role for one candidate. It does not scrape, rank, match, or aggregate jobs, and it never produces an unsupported fit percentage.
- No job or candidate ranking
- No salary or hidden-company inference
- No resume rewriting for keyword stuffing
- No proprietary Dataaxy marketplace logic
Dataaxy Academy
One repository, ready to share
Publish a recruiter-readable record of your prompt, evidence rules, structured output, evaluation, and limitations.