Python Resume Tailoring CLI
An in-progress Python CLI that assembles resume variants from validated evidence blocks and keeps every change visible through diff-based review.
What it is: A Python CLI for assembling role-specific resume variants from explicit evidence blocks without drifting into unsupported claims.
What I built: Designed the CLI workflow, evidence-block model, and review-first export flow for resume tailoring.
Current state: In progress: the implementation direction is real, but the system is still being built out.
Why it matters: Uses structured evidence blocks to avoid unsupported claims in generated variants.
Category: Tool / Infrastructure
Status: In Progress
Visibility: Public
What this project is
A Python CLI for assembling role-specific resume variants from a validated pool of evidence blocks.
Why I built it
Resume tailoring becomes repetitive fast, and generated variants drift easily when the evidence model is weak. I wanted a tooling layer that made the process faster while keeping the supporting evidence visible.
Constraints
- Output has to stay anchored to real evidence.
- Changes need to be reviewable before export.
- The workflow should help with consistency without turning into a black box.
Architecture
- Pydantic models for evidence blocks and output validation.
- Typer-based CLI commands for selecting roles, assembling variants, and reviewing diffs.
- Review-first export flow so the final resume version remains intentional.
Current state
This is actively in progress. The core direction is clear, but it still needs more iteration around templates and output ergonomics.
Why it matters
The useful part is not automated rewriting on its own. It is building a safer authoring loop for repetitive tailoring work where every change can still be checked before it becomes résumé truth.
Key decisions
- Use structured evidence blocks instead of free-form rewriting so generated variants stay defensible.
- Keep diff views in the loop so tailoring changes remain visible before export.
- Treat the CLI as a tooling layer, not as an automatic source of résumé truth.
What I'd improve next
The next improvement would be stronger template controls for different role families so structure can vary without weakening the evidence model.