Resume builder

Resume Builder

When to use

Any request to build, tailor, revise, or critique a resume.

Inputs

• Source material — read all that exist before drafting. $PERSONAL is the personal-docs sibling directory defined in CLAUDE.md; resolve it once at the start of the session.
  • $PERSONAL/career/impact-doc.md — always. Primary source of truth for roles, projects, and metrics.
  • $PERSONAL/career/brag-doc.md — read if it exists. Contains quantified wins and recognition not always captured in the impact doc. Cross-check metrics here before finalizing any bullet.
  • $PERSONAL/career/goals.md — read if it exists. Use to understand target role type, level, and what to emphasize or deprioritize.
  • $PERSONAL/career/personal-info.md — read if it exists. Use for the resume header: name, email, phone, location, LinkedIn, GitHub. If the file does not exist, use placeholders and note them explicitly so the user can fill them in.
  • $PERSONAL/career/current-resume.md — use as a baseline if it exists; otherwise build from scratch.
• Target (one of):
  • A specific job description → tailor to it
  • "General" → produce a strong default version
• Length preference: default 1 page. Confirm if I want longer.

Workflow

1. Confirm target

If the target isn't obvious from context, ask:

• Specific Job Description, or general?
• Length preference?
• Any role/company I'm asking you to deprioritize (e.g., I don't want my AWS work to dominate)?

2. Pull and rewrite bullets

For each role in my career, draft bullets using XYZ format:

Accomplished X by doing Y, measured by Z.

Every bullet must:

• Start with a strong action verb (built, led, shipped, reduced, architected — not "responsible for," "helped with," "worked on")
• Include scope (team size, system scale, dollar impact, user count, request volume — whatever applies)
• Include a quantified outcome where possible

If $PERSONAL/career/impact-doc.md is missing the numbers, surface that and ask me to fill them in rather than fabricating.

3. Tailor

When a target Job Description is provided:

• Match the Job Description's keyword vocabulary where it's truthful (e.g., if they say "distributed systems," use that phrase if it applies to my work, don't keyword-stuff with terms I haven't used)
• Reorder bullets so the most relevant come first within each role
• Cut bullets that don't reinforce the target — even strong ones
• Adjust the summary/header to mirror the role's framing

4. Critique pass

Before finalizing, evaluate every bullet:

• Does it have a number? If not, can I add one?
• Does it lead with action and outcome, or with the situation?
• Could a less impressive engineer write the same bullet? If yes, sharpen it.
• Is the scope clear?

Flag weak bullets explicitly with proposed alternatives.

5. Produce outputs

Create three formats in $PERSONAL/applications/<company>/<role>/ (or $PERSONAL/career/resumes/<variant>/ for general versions):

• resume.md — clean Markdown, the source of truth
• resume.docx — using the docx skill
• resume.pdf — using the pdf skill

For the .docx and .pdf, use a clean professional template — single column, sans-serif, no graphics, no skill bars, no photo. ATS-friendly.

6. Summarize the changes

End with a short summary of:

• What you emphasized and why
• What you cut and why
• Anything you flagged as weak that I should improve at the source (in $PERSONAL/career/impact-doc.md or $PERSONAL/career/brag-doc.md)

Anti-patterns

• Don't invent metrics. If the number isn't in my files, ask.
• Don't keyword-stuff. ATS optimization is real, but obvious stuffing reads worse than honest framing.
• Don't include an "Objective" section. Use a one-line professional summary instead, only if it adds signal.
• Don't list every technology I've ever touched. Curate to what matters for the target.
• One page unless explicitly asked otherwise. Senior-level resumes can justify two pages; if I'm targeting staff+, ask.
• Never use em dashes or en dashes (— or –) anywhere in the resume. Replace with a comma, colon, period, or parentheses. This applies to bullets, the summary, section headers, and job titles. No exceptions.
• Use generic, industry-standard terminology, not company-specific or tool-specific jargon. The reader may not share my team's vocabulary. Translate internal shorthand to widely understood equivalents before it hits the bullet. Examples:
  • "MR review cycle" → "code review cycle" (MR is GitLab-specific; PR is GitHub-specific; "code review" works everywhere)
  • "Cut MR review cycle by 50%" → "Cut code review cycle by 50%"
  • Internal product or team codenames → the public product name, or the generic capability ("checkout service," "payments pipeline")
  • Internal acronyms (CRQ, TPS, OKR-of-the-quarter names) → spelled-out plain English
  • Custom tooling names → the category of tool ("internal deploy tool" rather than the codename) If a term is genuinely industry-standard (Kubernetes, Kafka, gRPC, OAuth), keep it. If in doubt, ask whether a reader at a different company would recognize it.