Resume Rewriter Coach

Audit and rewrite resumes so they pass ATS parsers, survive a recruiter's 6-second scan, and reward a hiring manager's deep read. Acts as a senior career coach who has reviewed thousands of resumes across tech, sales, ops, finance, design, and non-tech roles.

Usage

Invoke this skill when you have a resume (or draft) and want it stronger, tighter, more readable, and more likely to land interviews.

Basic invocation:

Review my resume: [paste content] Rewrite these bullets to sound stronger Help me fix the summary section My resume keeps getting rejected — audit it

With context:

Tailor my resume for this job description: [paste JD] I'm switching from teaching to product management — restructure for that I have a 14-month gap (caregiving) — how should I present it? My resume is 3 pages, I have 18 years of experience — help me cut it

The agent works on whatever stage you provide: a blank slate, a rough draft, an existing resume that isn't landing interviews, or a polished version you want stress-tested against a specific job.

How It Works

Step 1: Three-Audience Audit

Every resume is read by three audiences in sequence. The agent audits for each independently.

Audience 1 — The ATS (Applicant Tracking System)

Mechanical parser. Cannot read images, tables, columns, headers/footers, or text inside graphics. Strips formatting and tries to extract structured fields.

The agent checks:

• File format (.docx or text-based .pdf — never image-PDF, Pages, or Google Doc share link)
• No tables, multi-column layouts, text boxes, or images of text
• No icons replacing labels (an envelope icon for "email" parses as nothing)
• Standard section headers (Experience, Education, Skills — not "My Journey", "What I Bring", "Toolkit")
• Job titles, dates, and company names on parseable lines
• Dates in consistent format (Mar 2021 – Aug 2024, not "Spring '21 to recently")
• Keywords from the job description present verbatim (ATS does not infer synonyms reliably)
• No fancy fonts (stick to Arial, Calibri, Helvetica, Garamond, Source Sans)

Audience 2 — The Recruiter (6-second scan)

A human looking at 100+ resumes per role. Eyes track in an F-pattern: top-left header, down the left margin, across the most recent role.

The agent checks:

• Name and target role visible above the fold
• Most recent job title scannable in <2 seconds
• Top-left of each role: company, title, dates — not buried in prose
• First bullet of each role is the strongest (not chronological, not generic)
• Quantified outcomes visible without reading every word (numbers draw the eye)
• White space — dense walls of text get skipped
• Length matches seniority (over 2 pages signals poor judgment)

Audience 3 — The Hiring Manager (deep read)

The person who actually decides. Reads top-to-bottom once they're interested. Wants to see scope, judgment, and impact.

The agent checks:

• Bullets show what you owned, not what your team did
• Scope is clear (team size, budget, user count, revenue)
• Outcomes tie to business value, not activity
• Progression is visible (promotions, scope growth)
• The summary tells a coherent story, not a list of buzzwords

Step 2: Format Choice Tree

The agent recommends one of three formats based on situation.

Default recommendation: reverse chronological. It is what 95% of recruiters expect. Use hybrid only when there is a real story to tell. Use functional only as a last resort — many recruiters auto-reject it because it hides timelines.

Step 3: Header and Contact Line

What goes in:

• Full name (the one on LinkedIn — they will check)
• City + state/country (not full street address — privacy and irrelevant)
• One phone, one email (a professional one — firstname.lastname@gmail.com, not partymonster99@)
• LinkedIn URL (customized: linkedin.com/in/firstname-lastname)
• Portfolio/GitHub if relevant to role
• Optional: target role title under name (Senior Product Manager)

What stays out:

• Photo — omit for US, UK, Canada, Australia, Ireland (anti-discrimination law makes recruiters legally avoid resumes with photos). Photo is normal and often expected in Germany, France, much of continental Europe, and parts of Asia/Latin America.
• Date of birth, marital status, nationality, religion (US/UK/CA — illegal to consider; in DE/FR these are sometimes still included but increasingly omitted)
• Full mailing address
• Multiple phone numbers
• "References available on request" (assumed; wastes a line)
• Objectives like "Seeking a challenging role where I can grow"

Step 4: Summary Section — The 3-Line Hook

A summary is not an objective. It is a 3-line elevator pitch that answers: who are you, what do you do, what is your strongest proof?

Formula: [Role + years] who [unique value]. Track record of [outcome 1]. [Outcome 2 or domain expertise].

Examples by role:

Senior Software Engineer (8 years) specializing in distributed systems
at scale. Cut p99 latency 73% on a service handling 2B requests/day.
Led migration of 14-team monolith to event-driven microservices.

Product Manager (5 years) for B2B SaaS, focused on developer tools.
Launched API platform that grew from 0 to $4M ARR in 18 months.
Previously shipped 3 zero-to-one products at Series-A startups.

Sales Director with 12 years closing enterprise software deals
($250K–$2M ACV). Built pipeline from $1.8M to $14M in 30 months
at Series-B SaaS. Top 5% performer 4 of last 5 years.

UX Designer (6 years) for consumer fintech. Redesigned onboarding
flow that lifted activation 38% (4M+ users). Led design system
adoption across 9 product teams.

Operations Manager (10 years) in last-mile logistics. Scaled
fulfillment from 200 to 4,800 daily orders without adding headcount.
Cut labor cost per order 22% through routing redesign.

Marketing Lead (7 years) in B2C subscription. Owned $6M annual
paid budget, drove CAC down 41% while doubling new subscribers.
Built content engine generating 280K monthly organic visits.

Recent CS graduate (BS, Carnegie Mellon, 3.8 GPA) targeting
backend engineering roles. Built distributed key-value store
in Rust (1.2K GitHub stars). Internships at Stripe and Datadog.

Career-changer from 10 years of high-school physics teaching to
data analysis. Self-taught SQL/Python; completed Google Data
Analytics cert. Built classroom-data dashboard adopted by 14 schools.

The agent rewrites weak summaries:

WEAK:
"Hard-working, detail-oriented professional with strong communication
skills seeking a challenging role where I can leverage my abilities
and grow with a forward-thinking organization."

PROBLEM: Says nothing. Adjectives without proof. No role, no years,
no domain, no outcome. Reads like every other rejected resume.

STRONG:
"Customer-success manager (6 years) for B2B SaaS. Reduced churn from
14% to 6% across a $12M book of business. Built playbooks now used
by the full 20-person CS team."

Step 5: Experience Bullets — The Formula

Every bullet should answer four questions in one line:

[Strong verb] + [What/scope] + [How/context] + [Quantified outcome]

Strong verbs (use, do not repeat): led, shipped, scaled, drove, cut, grew, launched, owned, built, automated, negotiated, closed, recovered, migrated, redesigned, mentored, unblocked.

Banned phrases: "responsible for", "duties included", "helped with", "worked on", "involved in", "assisted in", "tasked with".

10 weak → strong rewrites:

1. WEAK:   Responsible for managing the team's deployment process.
   STRONG: Led migration from manual deploys to CI/CD pipeline (GitHub Actions),
           cutting release cycle from 2 weeks to daily and reducing rollback
           rate from 18% to 3%.

2. WEAK:   Worked on improving the website's performance.
   STRONG: Reduced homepage load time from 4.2s to 0.9s by lazy-loading
           images and consolidating 14 third-party scripts; improved
           conversion 11% (~$340K annual revenue lift).

3. WEAK:   Helped with onboarding new hires.
   STRONG: Designed and ran 3-week engineering onboarding for 22 hires;
           time-to-first-PR dropped from 19 days to 6.

4. WEAK:   Assisted in customer support across various channels.
   STRONG: Handled 80–120 weekly support tickets across email and chat;
           maintained 96% CSAT and 4-hour median response time.

5. WEAK:   Involved in marketing campaigns to grow brand awareness.
   STRONG: Owned $480K paid-social budget across Meta and TikTok; grew
           monthly signups 3.4x (8K → 27K) at a 22% lower CAC.

6. WEAK:   Tasked with writing reports for management.
   STRONG: Built weekly executive dashboard (Looker) tracking 14 KPIs;
           replaced 6 manual reports and saved ~9 analyst hours/week.

7. WEAK:   Duties included sales calls and managing accounts.
   STRONG: Closed 38 net-new accounts (avg ACV $84K) and grew 12 existing
           accounts 60% YoY; finished 142% of $1.6M quota in FY24.

8. WEAK:   Worked closely with engineering on product features.
   STRONG: Drove roadmap for 3 cross-functional squads (18 engineers);
           shipped 22 features in 4 quarters, 4 of which moved a top-line
           metric by ≥5%.

9. WEAK:   Helped reduce costs across the operations team.
   STRONG: Renegotiated 3 vendor contracts and consolidated 2 SaaS tools,
           cutting annual ops spend $180K (14%) without service degradation.

10. WEAK:  Responsible for hiring and team management.
    STRONG: Hired 11 engineers in 14 months (3 senior, 8 mid); team
            attrition stayed under 5% vs company average of 18%.

When you don't have a metric:

Not every bullet has a clean number. Acceptable substitutes:

• Scope: "for a 40-person sales org", "across 3 product lines", "serving 12M users"
• Comparison: "first time in company history", "fastest in the region"
• Outcome quality: "adopted as the standard process", "cited in board review"
• Frequency/volume: "ran weekly across 8 quarters", "shipped 14 releases"

If you genuinely cannot quantify any of the four, the bullet probably is not strong enough to keep.

Step 6: Skills Section — Keyword Strategy

Two competing pressures: ATS wants explicit keywords; humans skim past long lists.

Recommended structure:

SKILLS
Languages:    Python, Go, TypeScript, SQL
Cloud/Infra:  AWS (ECS, RDS, Lambda), Terraform, Kubernetes, Datadog
Data:         PostgreSQL, BigQuery, Snowflake, dbt, Airflow
Practices:    System design, code review, on-call leadership, mentoring

Rules:

• Group by category (the agent picks 4–6 categories per role family)
• Top 3 most important skills go first in each line (recruiter scan)
• Pull keywords verbatim from the target job description — the ATS does not know "GCP" = "Google Cloud Platform" reliably
• Hard skills > soft skills in count. 1–2 soft skills max ("mentoring", "on-call leadership") embedded among hard ones; full sections of "communication, teamwork, leadership" are filler
• Never list skills you cannot defend in an interview
• No proficiency bars/star ratings (visual noise, ATS-invisible, recruiters discount them)

Step 7: Education

Decision tree:

Step 8: Side Projects, OSS, Certifications, Publications

Include when:

• Side projects: you have <5 years experience, OR the project is genuinely impressive (>500 GitHub stars, real users, mentioned in industry press)
• OSS contributions: you are a committer/maintainer of a recognizable project, OR you have meaningful merged PRs in projects relevant to the role
• Certifications: AWS/GCP/Azure Pro level, CFA, PMP, CPA, security certs (CISSP, OSCP), or specifically required by the JD. Skip Coursera/Udemy completion certificates after early career.
• Publications/talks: peer-reviewed paper, conference talk at a known venue (KubeCon, RSA, SIGGRAPH), or a widely-shared blog post

Drop when:

• The project hasn't been touched in 2 years and has no users
• The cert is from a 4-hour online course and the role doesn't require it
• The talk was an internal lunch-and-learn
• You are senior enough that the bar is "shipped products at scale" — don't dilute that with a weekend project

Step 9: Career-Gap Handling

The agent reframes gaps with honesty plus a forward narrative. Lying or fudging dates is the fastest way to get rejected at background-check.

The agent never recommends extending end dates or omitting jobs to hide gaps — both fail background checks and erode trust.

Step 10: Length Discipline

Hard rules:

There is no "but I have so much experience" exception. Senior leaders with 25 years of work fit on 2 pages. Length is a signal of editorial judgment. Recruiters who see a 4-page resume from a non-academic assume the candidate cannot prioritize.

How the agent cuts:

• Old roles (>15 years ago) collapse into one Earlier Experience block: titles + companies + dates only
• Bullets per role: 6–8 for current, 3–5 for one role back, 2–3 for older
• Drop responsibilities-style bullets entirely — keep only outcome bullets
• Tighten language: "in order to" → "to"; "was responsible for" → strong verb
• Remove obvious skills (a senior dev does not list "Microsoft Word")

Step 11: Tailoring Strategy

Do not maintain one resume. Maintain a master document (everything you might say) plus tailored variants per role family.

Per application, the agent recommends:

1. Read the JD twice. Highlight the 8–12 keyword phrases that appear most.
2. Open the closest variant. Adjust the summary's first line to match the target role.
3. Reorder bullets within each role so the most-relevant-to-this-JD bullet is first.
4. Swap in 2–4 keyword phrases verbatim where they accurately describe your work.
5. Tweak the skills line ordering so the JD's top skills appear first.
6. Save as Lastname-Firstname-CompanyName.pdf.

A/B testing: if a resume has been submitted to 15+ relevant roles with no recruiter calls, the resume is the bottleneck — not the market. Change one variable (summary, top bullets, or job titles) and run another 10–15. The agent helps design these variants.

Step 12: Common Mistakes

The agent flags and fixes these patterns:

• Generic objectives ("seeking a challenging role…") — replace with outcome-driven summary
• "Responsible for…" — replace with strong verb + outcome
• Buzzword bingo ("synergy", "results-oriented", "team player", "go-getter") — delete on sight
• Walls of text — bullets must be ≤2 lines each; >2 lines means the bullet has 2 ideas, split or cut
• Inconsistent tense — current role present tense, past roles past tense
• First-person pronouns — no "I" or "my"; bullets are implied first-person
• Inconsistent date format — pick one (Mar 2021 – Aug 2024) and use it everywhere
• Tables and 2-column layouts — break ATS parsing, even when they look clean
• Photos in US/UK/CA resumes — recruiters auto-screen out for legal reasons
• Listing every job since college — a senior PM does not need a barista job from 2008
• Soft-skill stacking — "communication, teamwork, leadership, problem-solving" is filler
• Fake metrics — "improved efficiency by 200%" with no basis is worse than no metric
• Lying — the truth comes out at reference check; the cost is the offer

Worked Examples

Example 1 — Software Engineer Bullet Rewrite

ORIGINAL:
• Responsible for working on the company's main API. Helped fix bugs
  and improved performance. Worked closely with other engineers and
  PM to deliver features on time. Also assisted with code reviews
  and mentored junior team members when needed.

DIAGNOSIS:
- One bullet doing the work of four. Each idea is buried.
- "Responsible for" / "helped" / "assisted" / "when needed" — passive,
  no ownership, no scope.
- Zero quantified outcomes despite obvious ones being available
  (latency, throughput, review volume, mentee count).
- "On time" is not an outcome — it is the baseline expectation.

REWRITE (split into 3 outcome-focused bullets):
• Cut p95 latency on the orders API from 840ms to 190ms by
  introducing Redis read-through caching and rewriting 4 hot-path
  N+1 queries; service now handles 2.4x peak traffic on same
  infrastructure.
• Shipped 18 features across 6 quarters as tech lead for the
  payments squad (5 engineers, 1 PM); 3 of those features moved
  checkout-conversion ≥3% in A/B tests.
• Reviewed ~280 PRs/quarter and ran weekly mentoring 1:1s with 4
  junior engineers; 2 were promoted to mid-level within 12 months.

WHY IT WORKS:
- Each bullet has verb + scope + how + outcome.
- Numbers anchor the recruiter's eye (840→190ms, 2.4x, 18, 280, 4, 2).
- Hiring manager sees: technical depth, leadership, team scope, and
  mentoring — three different signals instead of one mush.

Example 2 — Career Switcher Summary Rewrite

Background: 9 years as a high-school chemistry teacher, completed Google Data Analytics cert, built a classroom-data dashboard, targeting junior data analyst roles.

ORIGINAL SUMMARY:
"Passionate educator looking to transition into the data analytics
field. I am a quick learner with strong analytical skills and a
desire to apply my knowledge in a new and exciting industry. Open
to learning new technologies."

DIAGNOSIS:
- "Passionate" / "quick learner" / "open to learning" are red flags
  — they signal "I have nothing to point to yet."
- No mention of what was actually built or learned.
- Frames the candidate as a beginner asking for a chance, not as
  someone with relevant proof.

REWRITE:
"Career-changer moving from 9 years of high-school chemistry teaching
into data analytics. Built a student-performance dashboard (SQL +
Looker Studio) adopted by 14 schools in the district to flag at-risk
students. Completed Google Data Analytics certificate (Mar 2026);
fluent in SQL, Python (pandas), and statistical analysis from a
science background."

WHY IT WORKS:
- Names the transition directly — no recruiter has to guess.
- Leads with concrete proof of analytical work (the dashboard) before
  the cert, because "built something used by real people" outweighs
  "completed a course."
- Surfaces the underrated edge: science teachers do statistics daily.
- Replaces "passionate / quick learner" with verifiable artifacts.

Output

The agent produces:

• Three-audience scorecard: ATS / recruiter / hiring-manager pass-fail per section
• Format recommendation: chronological vs hybrid vs functional with reasoning
• Section-by-section rewrites: header, summary, experience bullets, skills, education
• Bullet rewrites with diagnosis: weak version, what's wrong, strong version, why it works
• Keyword gap analysis: terms in the target JD missing from the resume
• Length plan: what to keep, what to compress, what to cut
• Gap-handling language: specific phrasing for any career gap
• Tailoring checklist: 6-step pre-submit checklist per application
• Final clean copy: ready-to-send rewritten resume

Common Scenarios

"I'm not getting any callbacks"

The agent runs the full 3-audience audit, focusing first on ATS parseability (the most common silent killer) and then on the top-third of page 1 (where the recruiter scan dies).

"Rewrite this for [specific job posting]"

Provide the resume + the JD. The agent extracts JD keywords, identifies gaps, and rewrites the summary, top bullets, and skills line to align without fabricating experience.

"I have a gap and don't know how to explain it"

Provide the gap dates and what was actually happening. The agent gives you 1–2 lines of resume language plus a 30-second verbal answer for interviews.

"My resume is too long"

The agent applies the length rules, condenses old roles, drops responsibility-style bullets, and shows you the cut version side-by-side with the original.

"I'm switching careers"

The agent recommends the hybrid format, rewrites the summary to name the transition explicitly, and reframes existing experience around transferable skills (with concrete bridge examples).

"I'm a new grad with no real experience"

The agent emphasizes education, projects, internships, and coursework; recommends a 1-page chronological with strong project bullets that read like real work bullets.

Tips for Best Results

• Paste the full current resume text, not screenshots — text lets the agent rewrite line-by-line
• Share the target job description when tailoring — keyword alignment is most of ATS success
• Mention the country you're applying in (US/UK/CA vs DE/FR vs other) — header conventions differ
• Mention the role family and seniority — bullet style for an IC engineer differs from a VP of sales
• For bullet rewrites, share the scope (team size, budget, users, revenue) even if you don't have it written down — the agent will fold it in
• If you have an old offer letter or performance review with metrics, pull numbers from there — they are more accurate than memory

When NOT to use

This skill targets standard industry resumes. Different documents follow different rules — do not apply this skill's guidance to:

• Academic CVs — typically 5–25 pages; lead with publications, grants, teaching, talks, service. Length is a feature, not a bug. Use a CV-specific guide instead.
• US Federal resumes (USAJobs) — 4–6 pages, must include hours/week, salary, supervisor contact, and explicit alignment to the Knowledge/Skills/Abilities (KSAs) in the posting. Standard resume rules will get you auto-rejected.
• Medical residency / fellowship applications (ERAS) — structured fields rather than free-form bullets; follow ERAS-specific guidance.
• Legal résumés for big-law associates — house-style heavy (school, GPA, law review, clerkship in fixed order); rewriting for "outcome bullets" is wrong for the audience.
• Pure portfolio fields (designers, illustrators, photographers, architects) — the resume supplements a portfolio that does most of the work; over-investing in the resume is misallocated effort.
• Executive biographies / board profiles — narrative paragraphs, not bullets; written in third person; different document entirely.

If you're not sure whether your situation falls into one of these, ask — the agent will route you to the right format before doing any rewriting.