LinkedIn Job Resume Optimizer

Comprehensive job search automation with resume tailoring and interview preparation.

Overview

This skill orchestrates a complete job application workflow:

1. Search LinkedIn for relevant remote AI jobs
2. Extract full job descriptions and requirements
3. Analyze base resume to extract skills and experience
4. Generate ATS-optimized, job-tailored resumes
5. Perform skill gap analysis
6. Generate targeted interview preparation questions
7. Suggest LinkedIn profile optimizations

Workflow Diagram

User Request
    ↓
Workflow Orchestrator
    ↓
┌─────────────────┬──────────────────┬─────────────────┐
│ LinkedIn        │ Resume           │ ATS             │
│ Scraper         │ Analyzer         │ Optimizer       │
│ (Playwright)    │ (Pandoc+Docx)    │ (NLP)           │
└─────────────────┴──────────────────┴─────────────────┘
    ↓
Gap Analyzer (Compare skills)
    ↓
Question Generator (Interview prep)
    ↓
Final Report (Summary + File locations)

Prerequisites

1. Playwright MCP Server (for LinkedIn automation)

Start the server before running this skill:

bash /mnt/d/Ali_Home/Learning/AgenticAI/AI-P009/Assignments/ProjectA1/resume_optimizer/.claude/skills/browsing-with-playwright/scripts/start-server.sh

2. Base Resume

Ensure resume exists at configured path (default: /mnt/d/Ali_Home/Learning/AgenticAI/AI-P009/Assignments/ProjectA1/resume_optimizer/resume/resume06012026.docx)

3. System Dependencies

# Core tools
sudo apt-get install pandoc python3 python3-pip nodejs npm

# Optional (for PDF export)
sudo apt-get install libreoffice poppler-utils

4. Python Packages

pip install python-docx spacy nltk pandas requests
python -m spacy download en_core_web_sm
python -m nltk.downloader punkt stopwords

5. Node Packages

npm install -g @playwright/mcp

Quick Start

Basic Usage

python3 scripts/workflow_orchestrator.py \
  --job-keywords "AI Engineer remote" \
  --job-count 2

With Custom Resume Path

python3 scripts/workflow_orchestrator.py \
  --resume-path "/path/to/your/resume.docx" \
  --job-keywords "Machine Learning Engineer" \
  --job-count 2 \
  --output-dir "./custom_output"

Step-by-Step Workflow

Phase 1: Initialize Playwright Server

bash /mnt/d/Ali_Home/Learning/AgenticAI/AI-P009/Assignments/ProjectA1/resume_optimizer/.claude/skills/browsing-with-playwright/scripts/start-server.sh

Verify server is running:

curl http://localhost:8808

Phase 2: Search LinkedIn Jobs

python3 scripts/linkedin_scraper.py \
  --keywords "AI Engineer remote" \
  --count 2 \
  --output jobs.json

Output: jobs.json containing:

[
  {
    "title": "Senior AI Engineer",
    "company": "TechCorp",
    "url": "https://linkedin.com/jobs/view/12345",
    "description": "Full job description...",
    "required_skills": ["Python", "TensorFlow", "MLOps"],
    "preferred_skills": ["Kubernetes", "AWS"]
  }
]

Phase 3: Extract Base Resume Skills

# Convert DOCX to markdown
pandoc --track-changes=accept /mnt/d/Ali_Home/Learning/AgenticAI/AI-P009/Assignments/ProjectA1/resume_optimizer/resume/resume06012026.docx -o resume.md

# Extract skills
python3 scripts/resume_analyzer.py --resume-md resume.md --output base_skills.json

Output: base_skills.json with skills inventory

Phase 4: Generate Tailored Resumes

For each job, generate ATS-optimized resume:

python3 scripts/ats_optimizer.py \
  --base-resume /mnt/d/Ali_Home/Learning/AgenticAI/AI-P009/Assignments/ProjectA1/resume_optimizer/resume/resume06012026.docx \
  --job-description jobs.json \
  --job-index 0 \
  --output "resume_tailored_Senior_AI_Engineer.docx"

Phase 5: Perform Gap Analysis

python3 scripts/gap_analyzer.py \
  --base-skills base_skills.json \
  --job-requirements jobs.json \
  --output gap_analysis.json

Output: gap_analysis.json

[
  {
    "job_title": "Senior AI Engineer",
    "skill_gaps": [
      {
        "skill": "Kubernetes",
        "gap_type": "missing",
        "importance": "high",
        "related_experience": ["Docker experience in current role"]
      }
    ],
    "interview_questions": [
      "Explain your understanding of Kubernetes orchestration...",
      "How would you design a Kubernetes deployment for ML models?",
      "..."
    ],
    "linkedin_updates": {
      "about_section": "Add: 'Experienced in containerization with Docker, transitioning to Kubernetes for production ML deployments'",
      "skills_to_add": ["Kubernetes", "Container Orchestration", "MLOps"]
    }
  }
]

Phase 6: Generate Interview Questions

python3 scripts/question_generator.py \
  --gap-analysis gap_analysis.json \
  --questions-per-gap 10 \
  --output interview_prep.md

Phase 7: Cleanup

bash /mnt/d/Ali_Home/Learning/AgenticAI/AI-P009/Assignments/ProjectA1/resume_optimizer/.claude/skills/browsing-with-playwright/scripts/stop-server.sh

Configuration

Edit config.json in skill directory:

{
  "resume_path": "/mnt/d/Ali_Home/Learning/AgenticAI/AI-P009/Assignments/ProjectA1/resume_optimizer/resume/resume06012026.docx",
  "output_directory": "./resume_optimizer/output",
  "playwright_port": 8808,
  "job_search": {
    "default_keywords": "AI Engineer remote",
    "default_count": 2,
    "location": "Remote"
  },
  "ats_optimization": {
    "keyword_density_target": 0.75,
    "exact_match_priority": true,
    "preserve_formatting": true
  },
  "interview_prep": {
    "questions_per_gap": 10,
    "include_behavioral": true,
    "include_technical": true
  }
}

ATS Optimization Strategies

See references/ats-optimization-guide.md for detailed strategies including:

Keyword Density

• Required Skills: 70-80% coverage
• Preferred Skills: 40-50% coverage
• Exact Match: Use exact phrases from job description
• Context Integration: Weave keywords naturally into experience descriptions

Keyword Placement Priority

1. Professional Summary (highest ATS weight)

   • Include job title keyword
   • Top 3-5 required skills
   • Industry-specific terminology

2. Skills Section (exact match critical)

   • Mirror job posting's skills terminology exactly
   • Group by category if job posting does

3. Experience Descriptions (context + keywords)

   • Integrate keywords into achievement statements
   • Use action verbs + keyword + quantifiable result

Natural Integration Techniques

Action Verb + Keyword + Result:

Before: "Worked on machine learning projects"
After: "Developed TensorFlow-based ML models, reducing inference time by 30%"

Technology Stack Enumeration:

Before: "Built backend systems"
After: "Built scalable backend systems using Python, Docker, and Kubernetes on AWS infrastructure"

LinkedIn Automation

See references/linkedin-automation.md for:

• Navigation patterns for LinkedIn job search
• Element selectors and snapshot references
• Rate limiting and anti-bot detection strategies
• Fallback strategies if elements change

Key Points:

• Implement random delays (1-3 seconds between actions)
• Use --shared-browser-context flag for Playwright MCP
• Handle rate limiting with exponential backoff
• Parse snapshots for job card refs

Interview Question Generation

See references/interview-prep-patterns.md for:

• STAR method question templates
• Technical depth assessment questions
• Behavioral questions for skill gaps
• System design scenarios

Question Distribution:

• Missing Skills: 40% fundamentals, 30% transferable experience, 30% scenarios
• Weak Skills: 40% depth assessment, 30% practical application, 30% troubleshooting

Error Handling

No Jobs Found

if len(jobs) == 0:
    print("No jobs found matching criteria. Suggestions:")
    print("- Broaden search keywords")
    print("- Remove 'remote' filter")
    print("- Try alternative job titles")
    sys.exit(1)

Resume Read Errors

try:
    resume_text = extract_resume(resume_path)
except FileNotFoundError:
    print(f"Resume not found at {resume_path}")
    print("Please provide valid resume path with --resume-path")
    sys.exit(1)
except Exception as e:
    print(f"Error reading resume: {e}")
    print("Ensure resume is valid .docx format")
    sys.exit(1)

LinkedIn Rate Limiting

# Implement exponential backoff
for attempt in range(3):
    try:
        job_data = scrape_job(url)
        break
    except RateLimitError:
        wait_time = 2 ** attempt * 5  # 5s, 10s, 20s
        time.sleep(wait_time)

Playwright Connection Failures

• Pre-flight check: Verify server running before job search
• Auto-restart server if connection fails
• Retry with exponential backoff (3 attempts)

Troubleshooting

Issue	Solution
Playwright not responding	Restart: bash scripts/stop-server.sh && bash scripts/start-server.sh
LinkedIn blocks automation	Add wait times, use headless:false, rotate user agents
Resume conversion fails	Verify pandoc installed: pandoc --version
ATS optimization weak	Review keyword extraction, increase density target in config.json
No skill gaps detected	Lower similarity threshold in gap_analyzer.py
Python packages missing	Run: pip install python-docx spacy nltk pandas requests
spaCy model not found	Run: python -m spacy download en_core_web_sm

Output Summary Format

After running the workflow, you'll receive a summary report:

# Job Search Results Summary

## Jobs Found: 2

### Job 1: Senior AI Engineer - TechCorp
- **LinkedIn URL**: https://linkedin.com/jobs/view/12345
- **Tailored Resume**: ./resume_optimizer/output/resume_tailored_Senior_AI_Engineer.docx
- **Key Requirements**: Python, TensorFlow, MLOps, Kubernetes
- **Skill Gaps**: Kubernetes (High priority), AWS Sagemaker (Medium)
- **Interview Prep Questions**: 20 questions generated (see interview_prep.md)

### Job 2: Machine Learning Engineer - AI Startup
- **LinkedIn URL**: https://linkedin.com/jobs/view/67890
- **Tailored Resume**: ./resume_optimizer/output/resume_tailored_Machine_Learning_Engineer.docx
- **Key Requirements**: PyTorch, Docker, CI/CD, Model deployment
- **Skill Gaps**: PyTorch (High priority - similar TensorFlow experience)
- **Interview Prep Questions**: 15 questions generated (see interview_prep.md)

## LinkedIn Profile Recommendations

### About Section Updates
- Add: "Specialized in MLOps with focus on production-scale deployments"
- Emphasize: Kubernetes orchestration experience
- Highlight: Cross-functional collaboration in AI product development

### Skills to Add
1. Kubernetes (High priority - both jobs)
2. MLOps (Critical keyword)
3. Model Deployment
4. CI/CD for ML

## Next Steps
1. Review tailored resumes in ./resume_optimizer/output/
2. Study interview prep questions in interview_prep.md
3. Update LinkedIn profile per recommendations above
4. Apply to jobs with tailored resumes

Integration with Existing Skills

This skill leverages:

browsing-with-playwright

• LinkedIn job search automation
• Browser navigation and snapshot parsing
• Start server: bash scripts/start-server.sh
• Stop server: bash scripts/stop-server.sh

docx

• Resume reading using pandoc
• Tailored resume generation
• Convert: pandoc --track-changes=accept resume.docx -o resume.md

Advanced Usage

Manual Job URLs (Skip LinkedIn Scraping)

If you prefer to provide job URLs manually:

1. Create jobs_manual.json:

[
  {
    "title": "AI Engineer",
    "company": "CompanyName",
    "url": "https://linkedin.com/jobs/view/12345",
    "description": "Paste full job description here..."
  }
]

2. Run workflow with manual jobs:

python3 scripts/workflow_orchestrator.py \
  --manual-jobs jobs_manual.json \
  --skip-linkedin

Custom Keyword Extraction

Override automatic skill extraction:

python3 scripts/ats_optimizer.py \
  --base-resume resume.docx \
  --job-description jobs.json \
  --job-index 0 \
  --custom-keywords "Python,TensorFlow,Kubernetes,AWS" \
  --output tailored_resume.docx

Dependencies Summary

Required

• Python 3.8+
• Node.js 14+
• pandoc
• browsing-with-playwright skill
• docx skill

Python Packages

• python-docx
• spacy
• nltk
• pandas
• requests

Node Packages

• @playwright/mcp

Performance

• LinkedIn Search: 30-90 seconds (2 jobs)
• Resume Analysis: 10-20 seconds
• Resume Tailoring: 30-60 seconds per job
• Gap Analysis: 20-30 seconds
• Interview Prep: 40-60 seconds
• Total Estimated Time: 2-4 minutes for 2 jobs

Limitations

1. LinkedIn Access: Requires LinkedIn to be accessible (may need login for some content)
2. Job Count: Recommended maximum 5 jobs per session to avoid rate limiting
3. Resume Format: Only supports .docx format (not .doc or PDF)
4. Language: Optimized for English resumes and job descriptions
5. Technical Roles: Best suited for technical roles (AI, ML, Engineering)

Security & Privacy

• Local Processing: All resume processing happens locally
• No Storage: No resume data is stored remotely
• Browser Automation: Uses local Playwright instance
• Data Privacy: Job descriptions and resumes remain on your machine

License

This skill is part of the Claude Code skills ecosystem.