📹 THE FULL SCRIPT - Episode 02¶

Resume Fine-Tuner with Claude Sonnet 4.5¶

UPDATED: October 1, 2025 Claude Sonnet 4.5 Released: September 29, 2025

🎬 VIDEO TITLES (SEO-Optimized Options)¶

Primary Title (Recommended):

Claude Sonnet 4.5 Resume Agent: Get 10x More Employer Responses in 2 Minutes

Alternative Titles: 1. "Build Smart Resume AI with Claude Sonnet 4.5 | LaTeX + 100+ Templates Included" 2. "AI Resume Fine-Tuner: Claude Sonnet 4.5 - 10x More Job Offers in 2 Minutes" 3. "Stop Writing Resumes Manually - Claude Sonnet 4.5 Does It Better (Live Demo)" 4. "100+ Resumes, 10x More Offers: My Claude Sonnet 4.5 Resume Agent System"

SEO Keywords: Claude Sonnet 4.5, AI resume, resume agent, ChatGPT alternative, Gemini alternative, SWE-bench, job search automation, software engineer resume, LaTeX resume, Overleaf

⏱️ VIDEO STRUCTURE¶

Total Duration: 13-14 minutes (includes ~10 seconds of Veo 3 videos) Production Method:

2 on-camera segments (Hook + Closing) = 1-2 minutes
15+ screen recordings = 11-13 minutes (includes Claude Desktop usage demo)
3 Veo 3 videos (~10 seconds total) for professional transitions

Key Sections:

0:00-1:15: Hook (you on camera + LinkedIn/Gmail proof)
1:15-1:19: VEO 3 VIDEO #1 - Before/After transformation (4 sec)
1:19-3:01: Overleaf 100+ resumes + LaTeX templates
3:01-4:01: Why Claude Sonnet 4.5?
4:01-4:46: The Problem (job descriptions)
4:46-4:50: VEO 3 VIDEO #2 - System workflow (4 sec)
4:50-5:24: Agent explanation + diagram
5:24-6:54: Workspace setup + Claude Desktop token usage demo
6:54-8:24: Create the agent
8:24-10:54: Software Engineer Demo (LIVE, UNCUT)
10:54-11:54: LaTeX template reuse + Templates for other professions
11:54-12:24: Why Sonnet 4.5 wins
12:24-13:09: Iterative refinements
13:09-13:13: VEO 3 VIDEO #3 - Results celebration (4 sec)
13:13-14:02: Results & metrics
14:02-14:47: Closing + resources

🔥 HOOK (0:00-1:15)¶

Part 1: YOU ON CAMERA - Strong Open (5 seconds)¶

📹 YOU ON CAMERA - IMMEDIATE IMPACT

Camera Setup:

Phone or webcam at eye level
Face well-lit (window or ring light)
Clean background
Chest-up framing

Script to camera (POWERFUL, 5 SECONDS):

"10 times more job offers. 2-minute resumes. I'll prove it to you right now."

Recording Tips:

ONE TAKE, high energy, confident
Look directly at camera
Sharp, punchy delivery
Serious/determined look

Transition: IMMEDIATE CUT to LinkedIn (no fade, no pause)

📹 SCREEN: LinkedIn Messages & Gmail - Job Offers

What to show: 1. Open LinkedIn - Show your messages inbox

Scroll through multiple recruiter messages
Highlight: "Remote positions", "Senior roles", "Interview invitations"
Show at least 5-10 different companies reaching out
Blur company names if needed for privacy
Switch to Gmail - Show your inbox
Filter by "job" or "interview" or "offer"
Scroll through 10-15+ emails from recruiters
Show variety: startups, big tech, agencies
Highlight: Remote, salary ranges, "we'd love to chat"

Voiceover:

"10 times more job offers. 2-minute resumes. I'll prove it to you right now."

"Look at this. My LinkedIn. 

[Scroll through LinkedIn messages]

100 recruiter messages just this month. Remote positions. Senior roles.
Real opportunities.

[Switch to Gmail]

My Gmail. Same story. Interview requests. Salary ranges to be excited about.

[Pause]

This wasn't always the case. Let me show you what changed."

Editing Notes:

Blur sensitive info (company names, salaries if needed)
Add text overlay: "10+ messages this week"
Highlight "Remote" and "Senior" keywords
Keep it moving - don't linger on any one message

Part 3: You on Camera - The System (25 seconds)¶

📹 YOU ON CAMERA - Explanation

Script to camera:

"Claude Sonnet 4.5. Just released.
The best AI model for automation and coding.

[Lean forward slightly]

What you just saw? That's the result.

Those 100 recruiter messages? This is the tool behind them.

[Direct look to camera]

Here's the truth:

You won't be replaced by AI.
You'll be replaced by someone who knows how to USE it.

[Pause]

And I'm about to show you exactly how.

Let's build."

Recording Tips:

Conversational but confident
Quick pace, no rambling
Reference the LinkedIn/Gmail proof they just saw
End on "Let's build" with energy

Transition:

🎬 VEO 3 VIDEO #1 (3-4 seconds) - Before/After Transformation

Veo 3 Prompt: "An hourglass rotating. Left side shows sand taking 2 hours
to fall (labeled '2 HR'). Right side shows sand falling in 2 minutes
(labeled '2 MIN'). Smooth rotation between the two states. Dark background,
gold sand, elegant typography for numbers only."

Shows time savings visually
Sets up Overleaf proof
Then cut to Overleaf screen

📚 THE PROOF: 100+ RESUMES GENERATED (1:19-3:01)¶

📹 SCREEN: Overleaf - Your Resume Library (90 seconds)

What to show:

Part 1: The Volume (30 seconds) 1. Open Overleaf - Show your projects list

Scroll through 100+ resume projects
Show variety of filenames:
- resume_excelon_dotnet_fullstack_2024.tex
- resume_carnival_backend_senior_2024.tex
- resume_mufg_trading_platform_2024.tex
Keep scrolling to show the sheer volume

Part 2: LaTeX Template Selection (30 seconds) 2. Click "New Project" → Templates

Show Overleaf template gallery
Highlight professional templates:
- "Simple Hipster CV" ← Hover on this
- "Classic Resume"
- "Modern CV"
- "AltaCV"
Click on "Simple Hipster CV" to preview
Show how clean and professional it looks

Part 3: Open Real Example (30 seconds) 3. Open example resume - samples/example-latex-resume.tex

Show LaTeX source code on left
Show compiled PDF preview on right
Highlight key features:
- Professional formatting (automatic)
- Beautiful typography
- Perfect spacing and alignment
- ATS-friendly output
Scroll through to show:
- Different sections emphasized
- Specific keywords naturally integrated
- Project details tailored to job requirements

Voiceover:

"Before it, let me prove this works.

[Open Overleaf, start scrolling]

This is my Overleaf account. With 100 resumes.

[Keep scrolling]

Each one tailored to a specific job. .NET fullstack. Backend engineer.
Trading platform.

[Quick flash of template selection]

LaTeX format user here is optional but powerful - professional templates, beautiful formatting.
"Classic Resume"

     - "Modern CV"
     - "AltaCV"
       and hipster ones 

[Open example resume briefly]

Look at this. LaTeX source on the left, compiled PDF on the right.
Professional typography, perfect spacing, automatic formatting.

[Scroll through the PDF quickly]

Same experience. Different emphasis based on the job.
Carnival loyalty platform emphasized here because they wanted
microservices experience.

[Pause]

I've been doing this for months. The difference? Night and day.

After fine-tuning: it's 10x improvement in employer interest.

[Close Overleaf]

Now here's the key: You don't NEED LaTeX. Simple text file with your
experience works great. 

Claude generates the content. You choose the format.

Now, how do you build this system?"

Editing Notes:

Add counter overlay: "100+ LaTeX Resumes"
Speed up scrolling to 1.5x (show volume)
Zoom in on "Simple Hipster CV" template
Highlight split view: LaTeX source vs PDF output
Add "Beautiful PDFs, Any Template" text overlay
Add "10x More Responses" callout
Show LaTeX logo/badge in corner
Transition with: "Let me show you the secret..."

📊 WHY CLAUDE SONNET 4.5? (3:01-4:01)¶

Part 1: Model Comparison (45 seconds)¶

📹 SCREEN: Official Anthropic Models Page

Open: https://docs.claude.com/en/docs/about-claude/models/overview

Actions: 1. Scroll to model comparison table 2. Pause on each model (3 seconds each):

Opus 4.1 - Most powerful, expensive
Sonnet 4.5 ← ⭐ Highlight this with red box
Haiku 3.5 - Fastest, cheapest
Show pricing:
Sonnet 4.5: $3 input / $15 output per million tokens
Same price as old Sonnet 4!

Voiceover:

"Before we build anything, understand WHY Claude Sonnet 4.5.

Anthropic has three main models in their lineup:

- Opus 4.1: Extreme complexity, expensive - overkill for resumes
- Haiku 3.5: Ultra fast, cheap - but quality suffers
- Sonnet 4.5: The sweet spot - best for coding and agents

And here's the kicker: Same price as the old Sonnet 4.
Three dollars input, fifteen output per million tokens.
World-class AI, same cost."

Editing Notes:

Add red animated box around Sonnet 4.5 row
Zoom in on pricing column
Show "Same Price!" text overlay

Part 2: The Benchmarks That Matter (60 seconds)¶

📹 SCREEN: Anthropic Benchmark Announcement

Open: https://www.anthropic.com/news/claude-sonnet-4-5

Actions: 1. Scroll to benchmark section 2. Highlight each score (hold 5 seconds):

SWE-bench Verified: 77.2%
With parallel compute: 82.0%
GPT-5: 72.8% (show comparison)
Claude Opus 4.1: 74.5%

Voiceover:

"Why is Sonnet 4.5 the best? Numbers don't lie.

SWE-bench Verified: 77.2% - that's real-world coding problems.
Real GitHub issues. Real developer tasks.

[Pause while showing scores]

For context:

- GPT-5: 72.8%
- Claude Opus 4.1: 74.5%
- Claude Sonnet 4.5: 77.2%

And with extra compute? 82%.

This isn't marketing. This is why we use it for resume agents.
It understands complex instructions. It follows patterns.
It just works a bit better."

Editing Notes:

Animate numbers counting up
Add VS comparison graphic: Sonnet 4.5 vs GPT-5 vs Gemini
Green checkmark on Sonnet 4.5

Part 3: Live Claude Demo (45 seconds)¶

📹 SCREEN: Claude Code in Action

Record in Claude Code terminal:

Type this prompt:

Analyze this job description and extract:
1. Top 10 keywords to emphasize
2. Required vs. nice-to-have skills
3. Specific technologies mentioned
4. Experience level expectations

Job Description:
Senior Software Engineer - Cloud Infrastructure

- 7+ years backend development experience
- Expert in Kubernetes, AWS, microservices architecture
- Python or Go proficiency required
- Team leadership and mentoring experience
- Performance optimization at scale

Show: Claude's response streaming in real-time

Voiceover:

"Let me do a simple prove. Watch Claude Sonnet 4.5 analyze a job description.

[Silent for 15 seconds while Claude responds]

Look at this breakdown:

- Keywords extracted and prioritized
- Requirements categorized
- Context understood
- Technologies mapped

This is the brain of our resume agent. This is why it works."

Editing Notes:

DO NOT speed up Claude's response
Show real-time streaming
This builds trust

🎯 THE PROBLEM (4:01-4:46)¶

📹 SCREEN: Three Different LinkedIn Jobs (45 seconds)

Open 3 browser tabs: 1. linkedin.com/jobs/search/?keywords=senior%20software%20engineer 2. linkedin.com/jobs/search/?keywords=marketing%20manager 3. linkedin.com/jobs/search/?keywords=luxury%20real%20estate%20agent

Actions:

Quick tab switching (10 seconds each job)
Highlight different requirements on each:
Software: Python, AWS, Kubernetes, microservices
Marketing: SEO, analytics, content strategy, growth metrics
Real Estate: Sales volume, CRS/ABR certs, luxury market

Voiceover:

"Here's the core problem we're solving:

Your experience is broad and diverse.
Jobs are usually hyper-specific.

Software engineer job wants: Python, AWS, Kubernetes, microservices.
Marketing job wants: SEO, content strategy, analytics, growth metrics.
Real estate job wants: Luxury market, sales volume, specific certifications.

[Pause]

A generic resume? Gets ignored by ATS systems and recruiters.
A tailored resume? Gets you the interview.

But who has time to customize for EVERY job?
That's where Claude Code comes in."

Editing Notes:

Split screen showing all 3 jobs
Highlight keywords with colored boxes
Add "IGNORED" stamp on generic resume graphic
Add "INTERVIEW" checkmark on tailored resume

🤖 CLAUDE CODE AGENTS (4:46-5:20)¶

🎬 VEO 3 VIDEO #2 (3-4 seconds) - System Workflow

Veo 3 Prompt: "Three floating icons in sequence: 1. Document with magnifying
glass (analyzing), 2. Glowing AI chip (processing), 3. Perfect document with
checkmark (output). Icons appear one by one with smooth fade-in, connected by
light trails. Gradient background from dark blue to purple, cinematic lighting."

Shows system flow visually
Alternative to or supplement for Excalidraw
Then transition to detailed explanation

Part 1: What Are Agents? + System Diagram (90 seconds)¶

📹 SCREEN: Create Excalidraw Diagram (Live) OR Use Veo 3 Video Above

What to show: 1. Open Excalidraw in your browser 2. Create a simple workflow diagram showing: Job Description → Claude Agent → Tailored Resume ↓ ↑ Experience DB ────────────────┘

Or use Mermaid to generate, then import to Excalidraw: ![[Agent Tailored Resume.excalidraw]]

graph LR
    A[Job Description] --> B[Claude Agent]
    C[Experience Database] --> B
    B --> D[Keyword Matching]
    B --> E[Achievement Selection]
    D --> F[Tailored Resume]
    E --> F

Actions while drawing/showing: 1. Create boxes for: Job Description, Agent, Experience DB, Resume 2. Draw arrows showing data flow 3. Add labels: "Extract Keywords", "Match Experience", "Generate" 4. Keep it simple and clean

Voiceover:

"Quick context: What's an agent?

[Start drawing in Excalidraw or show pre-made Mermaid diagram]

Think of it like this: You have a job description here.
You have your experience database here.

The Claude agent sits in the middle. It:

- Reads the job description
- Extracts keywords and requirements
- Searches your experience for matches
- Selects the most relevant achievements
- Writes a tailored resume

[Point to each component as you mention it]

Now, I've covered agents in previous episodes - portfolio builders, automation tools.
Links below if you want those.

[Quick screen flash to Anthropic blog - 2 seconds]

Anthropic just released the Claude Agent SDK - the same infrastructure
powering Claude Code. Same building blocks for any agent.

Finance agent? Same SDK. Legal review? Same SDK. Customer support? Same SDK.

Us? Resume fine-tuning.

[Back to you or terminal]

Claude Code itself got new features:
Checkpoints, context editing, memory tool, Chrome extension.

Plus "Imagine with Claude" - their research preview.
Comment if you want a full episode on that."

Editing Notes:

Record Excalidraw drawing in real-time (sped up 2x)
OR show clean Mermaid diagram imported to Excalidraw
Add text overlays as you mention each capability
Add link graphic: "Previous Videos ↓ Description"
Smooth transition to GitHub resources

Part 2: Claude Desktop Token Tracking (30 seconds)¶

📹 SCREEN: Claude Desktop App - Usage Page

What to show: 1. Open Claude Desktop app 2. Navigate to Settings → Usage 3. Show usage bars clearly

Voiceover:

"One more thing: Token tracking is built into Claude Desktop now.

[Open Claude Desktop → Settings → Usage]

See this? 'Current session' - resets every 5 hours based on when you started.
Right now, 6% used, resets in 38 minutes.

'All models' - 24% used this week. Resets Monday 2 PM.
'Opus only' - 16% used. Separate tracking for premium model.

You need Claude Desktop installed for this. Cross-platform:
Mac, Windows, Linux - works everywhere.

Claude Code runs in terminal or VS Code - also cross-platform.

Now you know where your tokens go."

Editing Notes:

Zoom in on usage bars
Highlight each bar as explained
Add "5-Hour Window" overlay on Current session
Add "Weekly Limits" overlay
"Cross-Platform: Mac, Windows, Linux" badge
Show quick flash of VS Code with Claude running

📁 GITHUB RESOURCES & SETUP (5:24-6:54)¶

Part 1: GitHub Repository & Documentation (60 seconds)¶

📹 SCREEN: GitHub Repository

What to show: 1. Open: https://github.com/anton-abyzov/ai-power 2. Show repo structure - episodes folder, Episode 02 folder 3. Quick scroll through Episode 02 files 4. Open deployed docs: https://anton-abyzov.github.io/ai-power/

Voiceover:

"Important note on resources.

[Open GitHub repo]

Link to this GitHub repo is in the desc and it's public. Always will be

Every YouTube episode with detailed steps, prompts, code snippets,
important notes - all here.

[Scroll through Episode 02 folder]

Episode 02: Resume Fine-Tuner. Everything you'll see in this video.

[Open deployed docs site]

Don't have Obsidian? Don't want to mess with markdown files locally?

There's a deployed version with MkDocs. Browse tutorials right in your browser.

[Back to terminal/VS Code]

Everything's linked in the description. Let's build."

Editing Notes:

Show repo structure clearly
Highlight Episode 02 folder
Quick view of deployed docs
"All Free & Public" badge
GitHub star button animation

Part 3: Workspace Quick View (30 seconds)¶

📹 SCREEN: VS Code with Terminal

What to show: 1. VS Code open - workspace folder visible 2. Terminal panel - claude command running 3. Folder structure - quick view in sidebar

Voiceover:

"Back to the workspace. VS Code open, terminal ready.

[Show VS Code with terminal]

Four folders: experience-db for your career history, outputs for generated resumes,
job-descriptions to save postings, templates for base formats.

[Show folder structure in sidebar]

Simple structure. Let's build the agent."

Editing Notes:

Show VS Code with terminal and sidebar
Quick view of folder structure
Keep it moving

🎨 CREATE THE AGENT (6:54-8:24)¶

Part 1: Agent Configuration File (90 seconds)¶

📹 SCREEN: Create Agent in Terminal

Use terminal editor (nano/vim) or show file creation:

Create: resume-fine-tuner.md

Content:

# Resume Fine-Tuner Agent

## Description
Expert AI assistant for tailoring resumes to match job descriptions while maintaining authenticity and professional quality.

## Model
claude-sonnet-4-5-20250929

## Tools

- Read: Analyze job descriptions and experience databases
- Write: Generate tailored resume versions
- Grep: Search for specific skills and keywords
- Glob: Find relevant experience files
- Bash: Execute format conversions (Markdown → PDF/DOCX)

## Instructions

You are an expert resume writer and career strategist powered by Claude Sonnet 4.5, the world's best coding AI model (77.2% SWE-bench Verified).

### Primary Task:
Transform generic experience into job-specific resumes that pass ATS systems and impress recruiters.

### Workflow:
1. **READ** job description thoroughly
2. **EXTRACT** key requirements, technologies, and soft skills
3. **SEARCH** experience database for matching achievements
4. **PRIORITIZE** most relevant experiences
5. **WRITE** tailored resume emphasizing job-specific wins
6. **OPTIMIZE** for ATS keywords without keyword stuffing

### Critical Rules:

- **NEVER** fabricate companies I worked for 
- Use **realistic metrics** (20-40% improvements, not "10x" unless proven)
- **Vary language** naturally (developed, built, led, implemented, optimized)
- Maintain **professional tone** appropriate to industry
- Ensure **ATS compatibility** (clean formatting, keyword optimization)
- **Quantify** achievements wherever possible
- **Match job language** while staying authentic

### Output Requirements:

- **Format:** Professional, clean, ATS-friendly
- **Length:** 1-2 pages (strict maximum)
- **Sections:** Summary, Experience, Skills, Education, Certifications
- **Style:** Action verbs, bullet points, quantified results
- **Default output:** Markdown (easily convertible to DOCX/PDF)

### Quality Standards:

- 90%+ keyword match with job description
- Every bullet point starts with strong action verb
- Every achievement includes metric when possible
- No buzzwords without substance (e.g., "synergy", "rockstar")
- Professional formatting that works in all ATS systems

### Why Claude Sonnet 4.5?
Could you use ChatGPT-5, Gemini, or other models? Yes, they work.
But Claude Sonnet 4.5 offers:

- **Better instruction following** - Less hallucination, more accuracy
- **Superior keyword matching** - Contextual understanding, not just word stuffing
- **Smoother workflow** - More reliable, consistent results
- **77.2% SWE-bench score** - Proven coding and structured task excellence

This is why top professionals choose Claude Sonnet 4.5 for career automation.

Voiceover:

"Now we build the agent. Simple Markdown file.

I'm creating this in the terminal using nano - you can use any editor.

First: Basic metadata.

- Name: Resume Fine-Tuner Agent
- Model: claude-sonnet-4-5-20250929 - that's the exact model ID

Next: Tools it can use.

- Read: For job descriptions
- Write: For resumes
- Search: For matching keywords
- Bash: For format conversions

Then: The instructions. This is the brain.

[Pause while showing instructions]

Tell it what to do:
1. Read job description
2. Extract keywords
3. Search your experience
4. Match and prioritize
5. Write tailored resume

Rules to follow:

- Never fabricate companies I worked for
- Use realistic metrics
- Maintain authenticity
- Optimize for ATS

And finally, output format:

- Clean, professional
- 1-2 pages max
- Bullet points with metrics
- ATS-friendly

[Pause]

Now, could you use ChatGPT or Gemini? Sure.
But Claude Sonnet 4.5 is smoother. More accurate.
Better at complex instructions. 77.2% SWE-bench proves it.

Save the file. Agent created."

Editing Notes:

Speed up file creation to 1.5x
Pause on "Model" line - zoom in
Highlight key sections as voiceover mentions them
Add "77.2% SWE-bench" badge in corner

💼 DEMO 1: SOFTWARE ENGINEER (7:24-9:54)¶

Part 1: Create Experience Database (60 seconds)¶

📹 SCREEN: Create Software Engineer Experience File

Create: experience-db/software-engineer.md

Content:

# Software Engineering Experience Database

## Current Role: Senior Backend Engineer @ TechCorp
**Period:** January 2023 - Present
**Location:** San Francisco, CA (Remote)
**Tech Stack:** Python, FastAPI, PostgreSQL, Redis, Docker, Kubernetes, AWS
**Team Size:** Led team of 5 engineers
**Scale:** 10M+ requests per day, 99.9% uptime SLA

### Key Achievements:

- **Performance Optimization:** Reduced API response latency by 28% through Redis caching strategy and query optimization
- **Microservices Migration:** Led migration from monolithic architecture to 12 microservices, improving deployment speed by 45%
- **Event-Driven Architecture:** Implemented Kafka event streaming handling 500K+ events/day with zero message loss
- **Mentorship:** Mentored 2 junior engineers to mid-level positions within 8 months
- **Reliability:** Maintained 99.9% uptime across all services during high-traffic periods (Black Friday, product launches)
- **Cost Optimization:** Reduced AWS infrastructure costs by 22% through rightsizing and reserved instances

## Previous: Full Stack Developer @ StartupXYZ
**Period:** March 2021 - December 2022
**Location:** Austin, TX
**Tech Stack:** React, Node.js, MongoDB, AWS, Docker
**Product:** SaaS platform serving 50,000+ active users

### Key Achievements:

- **Authentication System:** Built OAuth2 authentication supporting Google, GitHub, and email
- **Performance:** Reduced initial page load time by 35% through code splitting and lazy loading
- **Real-Time Features:** Implemented WebSocket-based real-time collaboration features
- **CI/CD:** Established CI/CD pipeline with GitHub Actions, reducing deployment time from 2 hours to 15 minutes
- **Testing:** Increased test coverage from 40% to 85%, reducing production bugs by 60%

## Previous: Junior Developer @ AgencyABC
**Period:** June 2020 - February 2021
**Tech Stack:** JavaScript, React, Node.js, MySQL

### Key Achievements:

- Developed 15+ client websites using React and Node.js
- Implemented responsive designs increasing mobile engagement by 40%
- Reduced page load times by 25% through image optimization

## Education
**BS in Computer Science** - University of California, Berkeley (2020)
**GPA:** 3.7/4.0
**Relevant Coursework:** Data Structures, Algorithms, Distributed Systems, Machine Learning

## Certifications

- **AWS Certified Solutions Architect** - Associate (2023)
- **Kubernetes CKA** - Certified Kubernetes Administrator (2024)
- **MongoDB Certified Developer** (2022)

## Technical Skills

### Languages (Proficiency: Expert → Intermediate)

- **Expert:** Python, JavaScript, SQL
- **Advanced:** Go, TypeScript
- **Intermediate:** Java, Rust

### Frameworks & Libraries

- **Backend:** FastAPI, Django, Node.js, Express
- **Frontend:** React, Next.js, Vue.js
- **Testing:** Pytest, Jest, Cypress

### Cloud & Infrastructure

- **AWS:** EC2, S3, RDS, Lambda, ECS, CloudWatch
- **Kubernetes:** Deployment, Services, Ingress, Helm
- **Docker:** Multi-stage builds, Docker Compose
- **CI/CD:** GitHub Actions, Jenkins, GitLab CI

### Databases

- **SQL:** PostgreSQL, MySQL
- **NoSQL:** MongoDB, Redis, DynamoDB
- **Message Queues:** Kafka, RabbitMQ

### Tools & Practices

- **Version Control:** Git, GitHub, GitLab
- **Monitoring:** Grafana, Prometheus, Datadog
- **Agile:** Scrum, Kanban, Jira
- **Code Review:** Strong peer review culture, 100+ PRs reviewed

## Notable Projects
1. **Payment Processing System** - Handled $10M+ in transactions with PCI compliance
2. **Analytics Dashboard** - Real-time metrics dashboard serving 1000+ concurrent users
3. **API Gateway** - Built centralized API gateway handling 50M+ requests/month

## Soft Skills

- Technical leadership and mentoring
- Cross-functional collaboration (Product, Design, QA)
- Clear technical communication to non-technical stakeholders
- Agile/Scrum methodologies
- Remote team collaboration

Voiceover:

"Demo - Software engineer resume.

First step: Create your experience database. This is YOUR career, structured.

Start with current role:

- Company, dates, tech stack
- Team size, scale metrics
- Every achievement with numbers

[Pause showing achievements]

Notice the metrics: 28% latency reduction, not "10x faster"
45% deployment improvement, not "revolutionary speedup"
Real numbers. Defensible in interviews.

Previous roles: Same structure.
Technologies, achievements, metrics.

Education, certifications, skills - everything categorized.

This is your ammunition. Claude will match it to jobs.
The more detailed, the better the output."

Editing Notes:

Speed up file creation 2x
Pause on achievements section
Highlight metrics with colored boxes
Add "Real Numbers = Interview-Proof" text overlay

Part 2: Find Real Job (30 seconds)¶

📹 SCREEN: LinkedIn Job Search

Actions: 1. Open: linkedin.com/jobs/search/?keywords=senior%20backend%20engineer%20python 2. Find real job posting (e.g., Stripe, Airbnb, or mid-size tech company) 3. Show job description 4. Highlight key requirements:

7+ years Python
Kubernetes and AWS
Microservices experience
Team leadership
Copy full job description

Voiceover:

"Real job. LinkedIn. Senior Backend Engineer at ClickUp.

Requirements they're asking for:

- 5+ years Node.js experience
- High quality in Typescript
- AWS expertise
- Product-minded

Salary: $180K-$220K. Remote.

Perfect test case. Let's tailor the resume."

Editing Notes:

Blur company logo if needed for privacy
Highlight requirements with colored underlines
Show "Real Job, Real Test" badge

Part 3: Show the Result - LaTeX Resume Example (60 seconds)¶

📹 SCREEN: Open LaTeX Resume from Repo

Actions: 1. Navigate to samples/example-latex-resume.tex in repo 2. Scroll through showing the perfectly tailored .NET resume 3. Briefly open compiled PDF version to show final output

Voiceover:

"This is what the agent produces. Here's a real example from the repo.

[Open example-latex-resume.tex]

Look at this. Tailored for a .NET Fullstack Developer position.

[Scroll through slowly]

Professional summary - hits all keywords:
'18+ years building enterprise applications using C# .NET, Angular, SQL Server...'
'OAuth/OIDC authentication, REST API design, T-SQL optimization...'

Experience section - client projects with generic company names:

- Entertainment Platform Company - Angular 15, OAuth 2.0, microservices
- Financial Services Platform - Trading APIs, sub-10ms response times
- Healthcare Data Platform - HIPAA-compliant, 10M+ medical events daily
- Sports Analytics Platform - Real-time processing, 200K+ messages/second
- Healthcare Management System - Practice integration, API-to-API communication

Every bullet has metrics. Every achievement is quantified.
150K+ requests/second. 50M+ records daily. Team of 12 developers.

[Scroll to skills section]

Technical skills perfectly organized:

- Languages: C#, TypeScript, T-SQL
- Frameworks: .NET Core 6-8, Angular 14-17, Entity Framework
- Web Tech: REST APIs, OAuth 2.0, OIDC, JWT
- Databases: MS SQL Server, T-SQL optimization

[Quick flash of PDF version]

This compiles to a beautiful PDF. Professional. ATS-friendly.

The agent does this. Reads job description. Searches your experience database.
Writes tailored resume. 90%+ keyword match. 2 minutes.

You customize the template for your profession. Same principle.
Marketing? Real estate? Software? Doesn't matter. Agent adapts."

Editing Notes:

Scroll slowly through LaTeX file to show structure
Highlight key metrics and keywords
Quick 3-second flash of compiled PDF
Add "Example Output" badge
"Available in Repo" text overlay

🎨 OUTPUT FORMATS & OTHER PROFESSIONS (9:54-10:54)¶

Part 1: Output Formats - Your Choice (30 seconds)¶

📹 SCREEN: Show Multiple Format Options

What to show: 1. Generated Markdown resume - outputs/senior-backend-engineer-stripe-2025-10-01.md 2. Quick conversion options:

Copy to Overleaf → LaTeX PDF (show example in repo: samples/example-latex-resume.tex)
Save as .docx (Claude Code can write Word-compatible markdown)
Keep as plain text for ATS systems
Open example LaTeX resume from repo briefly

Voiceover:

"Output format? Your choice.

[Show generated MD file]

Claude generates Markdown by default. Clean, ATS-friendly.

[Show repo samples folder]

Want beautiful LaTeX PDFs? Example in the repo - 'example-latex-resume.tex'.
Copy to Overleaf, paste content, compile. Professional output.

[Quick flash of LaTeX PDF]

Want Word format? Claude Code can write directly to .docx-compatible markdown.
Want plain text? You already have it.

The content is perfect. The format? Whatever you need.

That's why I have 100+ resumes. Same agent, any format, 5 minutes each."

Editing Notes:

Show outputs folder with .md file
Quick cut to repo samples/ folder
Flash LaTeX PDF example (2 seconds)
"Any Format You Need" text overlay
"Examples in Repo" badge

Part 2: Templates for Other Professions - DIY (30 seconds)¶

📹 SCREEN: GitHub Repo - Templates Folder

What to show: 1. Open GitHub repo in browser 2. Navigate to templates/ folder 3. Show available templates:

experience-software-engineer.md ✅ (we just did this live)
experience-marketing-manager.md
experience-real-estate-agent.md
(scroll to show more if you have them)
Quick preview of Marketing template (10 seconds)
Quick preview of Real Estate template (10 seconds)

Voiceover:

"What about other professions? Marketing? Real estate? Sales?

[Open GitHub templates folder]

I've included templates for you. Same structure, different professions.

[Show marketing-manager.md]

Marketing manager: SEO, content, analytics, growth metrics.
All the channels, tools, certifications you'd need.

[Show real-estate-agent.md]

Real estate agent: Sales volume, certifications, client ratings.
Completely different world, same principle.

[Pause]

The process? Exactly what we just did with software engineering.

1. Copy the template for your profession
2. Fill in YOUR experience
3. Use the agent with any job description
4. Get tailored resume in 2 minutes

Marketing, sales, real estate, finance, healthcare - whatever.
The system adapts. The agent works.

I'm not going to do live demos for each profession.
You got the idea. Templates are there. Adjust for your needs. Go for it."

Editing Notes:

Show templates folder clearly
Quick scroll through each template
Add "DIY - Use These Templates" badge
"All Professions Supported" overlay
GitHub star button highlight

Part 1: Quick Adjustments - Conversational Editing (45 seconds)¶

📹 SCREEN: Claude Code - Iterative Refinements

What to show: 1. Generated resume open in editor 2. Chat with Claude to make adjustments 3. Show examples:

"Make the summary more concise"
"Add more metrics to the first project"
"Emphasize leadership more"
"Tone down the technical jargon"

Type in Claude:

@resume-fine-tuner

The resume looks great, but can you:
1. Make the professional summary shorter (2 sentences max)
2. Add more quantified metrics to the Carnival project
3. Emphasize team leadership throughout

Keep everything else the same.

Show Claude making adjustments in real-time

Voiceover:

"Best part? Iterative refinement.

[Show chatting with Claude]

Resume generated, but you want tweaks?

'Make the summary shorter.'
'Add more metrics here.'
'Emphasize leadership more.'

[Show Claude updating]

Just tell Claude what to adjust. It remembers the context,
makes the changes, keeps everything else intact.

No starting over. No manual edits. Just conversation.

Perfect resume in 2 minutes. Fine-tuned in 30 seconds more."

Editing Notes:

Show chat interface clearly
Highlight changes being made
Add "Iterative Refinement" text overlay
Show before/after of adjustments
"Conversational Editing" badge

📈 RESULTS & THE BIG PICTURE (11:24-12:13)¶

🎬 VEO 3 VIDEO #3 (3-4 seconds) - Results Celebration

Veo 3 Prompt: "Numbers appearing and growing in 3D space: '10x' appears first
(large, gold, rotating), '100+' appears next (silver, floating), '2 min' appears
last (bronze, pulsing). Confetti particles falling in background. Professional
office setting, warm lighting, success theme colors."

Why this works:

Shows success metrics visually
Celebratory feel before results discussion
Only uses numbers (no spelling risk)
Sets positive tone for metrics section

Alternative (simpler):

Veo 3 Prompt: "A graph line shooting upward from bottom left to top right.
Starting point labeled 'BEFORE', ending point labeled 'AFTER'. Green line with
glowing trail effect, particles following the path. Dark background with grid,
stock market success aesthetic."

Transition: Cut from Veo 3 video to metrics comparison screen

Part 1: Metrics That Matter (30 seconds)¶

📹 SCREEN: Before/After Comparison

Show visual comparison:

BEFORE (Generic Resume):
├── Applications sent: 50
├── Response rate: 12% (6 responses)
├── Interviews: 3
├── Time per application: 90 minutes
└── Total time invested: 75 hours

AFTER (Claude-Tailored Resumes):
├── Applications sent: 50
├── Response rate: 44% (22 responses)
├── Interviews: 18
├── Time per application: 3 minutes
└── Total time invested: 2.5 hours

IMPROVEMENT:
├── Response rate: +267%
├── Interviews: +500%
├── Time saved: 72.5 hours
└── ROI: Massive

Voiceover:

"Real results. Real numbers.

Before automation:

- 50 applications
- 12% response rate (6 responses)
- 3 interviews
- 90 minutes per application
- 75 hours total

After Claude Sonnet 4.5:

- 50 applications
- 44% response rate (22 responses)
- 18 interviews
- 3 minutes per application
- 2.5 hours total

[Pause]

That's 3.7x more responses.
6x more interviews.
72 hours saved.

This isn't theory. This is math."

Editing Notes:

Animated counters
Before (red) vs After (green) split screen
Confetti animation on improvements
"ROI: Massive" with trophy

Part 2: The Real Message (30 seconds)¶

📹 SCREEN: Text overlay with subtle background

Show text appearing:

"You won't be replaced by AI.

You'll be replaced by someone who knows how to USE AI.

The difference?
This video.
This agent.
2 minutes of automation.

You now have the knowledge.
You now have the tools.

The question is: Will you use them?"

Voiceover:

"Let's zoom out. The bigger picture.

You won't be replaced by AI.
But you might be replaced by someone who leverages AI better than you.

[Pause]

Think about it:

- They apply to 50 jobs in 3 hours
- You apply to 10 jobs in 15 hours
- They get 20+ interviews
- You get 2-3

Same experience. Same qualifications.
The only difference? They built this agent.

[Pause]

You now have everything you need:

- The best AI model (Claude Sonnet 4.5)
- The complete agent setup
- Templates for 3+ professions
- The exact prompts

Everything is in the description. Free. Public. Yours.

The question is: Will you use it?"

Editing Notes:

Dramatic text reveals
Subtle inspiring background music
"Your Move" final text
Pause for impact

🎬 CLOSING (12:13-12:58)¶

📹 YOU ON CAMERA (45 seconds)

Camera setup: Same as opening

Script:

"You just watched how to build a resume fine-tuning system with
Claude Sonnet 4.5 - the world's best coding model.

[Pause]

Not magic. Not hype.
The world's best coding model, released three days ago,
applied to your career.

[Look directly at camera]

Everything you saw is free on GitHub. Link in description.

- Agent configuration files
- Experience templates (software, marketing, real estate, and more)
- LaTeX examples
- All the prompts
- Full documentation

[Lean forward, confident]

Here's your choice:

- Spend 2 hours customizing each resume manually
- Or spend 2 minutes with Claude Sonnet 4.5

One approach gets you ignored.
The other gets you interviews.

[Pause, slight smile]

The people who win in the AI age aren't the ones with the best raw skills.
They're the ones who know how to amplify those skills with AI.

You now have the blueprint.

[Point to camera]

Go build. Get interviews. Get offers.
And when you do, comment below and tell me about it.

Let's go."

Recording tips:

High energy, inspiring tone
Make eye contact with camera
Confident body language
End with clear call-to-action

END SCREEN (20 seconds)¶

📹 SCREEN: YouTube End Screen Template

Elements:

Big Subscribe Button (center)
Previous Video: Episode 01 - Portfolio Builder (left)
GitHub Link: github.com/anton-abyzov/ai-power (right)
Your Channel Logo/Face (top)

Voiceover OR Text Overlay:

"Subscribe for more AI automation tutorials.
Check out Episode 01: No-Code Portfolio Builder.

And if you want to see Claude Imagine - Anthropic's new feature -
comment below. I'll do a full episode on it.

Drop your results in the comments - I read every one.

Now go build something incredible. Do it now."

Music: Upbeat, inspiring (fade out)

📊 PRODUCTION SUMMARY¶

What You'll Record:¶

On-Camera (60 minutes recording time):

Opening hook: 45 seconds (record 3-5 takes)
Closing: 45 seconds (record 3-5 takes)
Total on-camera: ~1.5 minutes final footage

Screen Recordings (4-5 hours): 1. Model comparison pages (2 recordings, 5 min) 2. Benchmark pages (2 recordings, 5 min) 3. Claude live demo (1 recording, 5 min) 4. LinkedIn job searches (3 recordings, 10 min) 5. Terminal setup (2 recordings, 5 min) 6. Agent creation (1 recording, 10 min) 7. Experience databases (3 recordings, 15 min) 8. Claude agent demos (3 recordings, 30 min - UNCUT!) 9. Resume reviews (3 recordings, 10 min) 10. Format conversions (1 recording, 5 min) 11. GitHub tour (1 recording, 5 min)

Voiceover Recording (2 hours):

Record in quiet room with good mic
Follow script closely but stay natural
Record in sections for easier editing
High energy throughout

Editing (6-8 hours):

Sync voiceover to screen recordings
Add text overlays for keywords
Animate comparison tables
Add progress indicators on Claude demos
Color grade for professional look
Add subtle background music (-28dB)
Create custom thumbnail

Total Production Time: 14-18 hours

✅ PRE-PRODUCTION CHECKLIST¶

Before Recording:¶

Technical Setup:

[ ] Claude Sonnet 4.5 access confirmed
[ ] Claude Code installed and working from terminal
[ ] OBS Studio configured for screen recording
[ ] Microphone tested (USB mic or quality headset)
[ ] Camera setup for on-camera segments (phone or webcam)
[ ] Good lighting for on-camera (ring light or window)

Environment:

[ ] Desktop cleaned (no personal files visible)
[ ] Notifications disabled (macOS Do Not Disturb / Windows Focus Assist)
[ ] Browser bookmarks bar cleaned
[ ] Terminal font size: 18-22pt (readable on recording)
[ ] IDE/Editor font: 16-18pt
[ ] Browser zoom: 125% (easier to see)
[ ] Dark mode enabled everywhere

Content Preparation:

[ ] 3 real job descriptions found and saved:
Software Engineer (Python/AWS/Kubernetes)
Marketing Manager (SaaS/SEO/Content)
Real Estate Agent (Luxury/High-volume)
[ ] Agent configuration file drafted
[ ] Experience database templates customized
[ ] All URLs bookmarked and tested

Files & Folders:

[ ] Workspace folder created: resume-workspace/
[ ] Subfolders created: templates/, outputs/, experience-db/, job-descriptions/
[ ] Agent file ready: resume-fine-tuner.md

🎯 CRITICAL SUCCESS FACTORS¶

1. Show Real-Time Claude Demos (MOST IMPORTANT!)¶

DO NOT speed up or cut Claude's responses
Show the full thinking process (60-90 seconds)
This is your proof - builds massive trust
Viewers need to see it actually works

2. High Energy On Camera¶

20% louder and faster than normal conversation
Smile while talking (sounds friendlier in voiceover)
Confident body language
You're teaching something powerful - show excitement!

3. Use Real Jobs¶

No fake or made-up job descriptions
Find actual postings on LinkedIn
Shows it's a real-world solution
Builds credibility instantly

4. Emphasize Claude Sonnet 4.5¶

Mention "Claude Sonnet 4.5" at least 10 times
Show the 77.2% SWE-bench score multiple times
Compare to ChatGPT and Gemini explicitly
This is your SEO and differentiation

5. Three Different Professions¶

Software engineer (tech)
Marketing manager (creative/analytical)
Real estate agent (sales/non-tech)
Shows universal applicability
Broader audience appeal

6. The "AI Replacement" Message¶

"You won't be replaced by AI..."
"...but by someone who uses AI better"
Inspirational and empowering
Calls to action

🚨 COMMON MISTAKES TO AVOID¶

❌ Speeding up Claude responses → Ruins the "real" proof ❌ Low energy on camera → Boring, viewers leave ❌ Using fake job descriptions → Kills credibility ❌ Forgetting to show model selection → Misses key differentiation ❌ Not testing agent before recording → Errors on camera, waste time ❌ Poor audio quality → Viewers can't understand, leave immediately ❌ Cluttered screen → Looks unprofessional, distracting ❌ No clear call-to-action → Viewers don't know what to do next

📈 POST-PRODUCTION & UPLOAD¶

YouTube Upload Checklist:¶

Video Details:

Title: "Build AI Resume Agent with Claude Sonnet 4.5 (77.2% SWE-bench - Beat ChatGPT & Gemini!)"
Description:

Claude Sonnet 4.5 just dropped with a 77.2% SWE-bench score - the best coding AI model ever.

I'm showing you how to build an AI Resume Agent that creates perfectly tailored resumes in under 2 minutes.

🔗 RESOURCES (All Free):

- GitHub Repo: [your-link]
- Full Documentation: [mkdocs-link]
- Episode 01 (Portfolio Builder): [link]

⏱️ TIMESTAMPS:
0:00 - Introduction
0:45 - Why Claude Sonnet 4.5?
3:00 - The Resume Problem
3:45 - Claude Agent SDK
5:00 - Workspace Setup
6:00 - Create the Agent
7:30 - Demo 1: Software Engineer
10:00 - Demo 2: Marketing Manager
12:00 - Demo 3: Real Estate Agent
14:00 - Why Sonnet 4.5 Wins
14:45 - Output Formats
15:15 - Results & Metrics
16:00 - Closing & Next Steps

🎯 WHAT YOU'LL LEARN:
✅ Build AI resume agent with Claude Sonnet 4.5
✅ 3x higher interview response rate
✅ 60x faster applications (2 min vs 2 hours)
✅ Works for ANY profession (software, marketing, real estate, etc.)
✅ Complete templates and prompts included

💡 KEY FEATURES:

- Claude Sonnet 4.5 (77.2% SWE-bench score)
- Beats ChatGPT-5 and Gemini
- Lower hallucination, better accuracy
- Same price as old Sonnet 4

🚀 GET STARTED:
1. Clone the GitHub repo
2. Install Claude Code
3. Copy the agent configuration
4. Create your experience database
5. Generate tailored resumes in minutes

You won't be replaced by AI. You'll be replaced by someone who knows how to USE AI.

This video gives you the blueprint. Now go build.

#ClaudeSonnet45 #AIResume #JobSearch #CareerAutomation #ClaudeCode #AIAgent #SoftwareEngineering #Marketing #RealEstate

Tags (Max 500 characters):

Claude Sonnet 4.5, AI resume, resume automation, ChatGPT alternative, Gemini alternative, SWE-bench, job search, career automation, AI agent, Claude Code, software engineer resume, marketing resume, real estate resume, ATS optimization, resume builder

Thumbnail Design:

Your face (excited/confident expression)
Text: "Claude Sonnet 4.5"
Badge: "77.2% SWE-bench"
Subtext: "3 Industries, 2 Minutes Each"
High contrast colors (yellow/black or blue/white)
Large readable fonts

Playlist:

Add to "AI Power" series
Create if doesn't exist

Cards:

3:00 - Episode 01 card
14:00 - GitHub repo card
Subscribe reminder card

🎯 SUCCESS METRICS TO TRACK¶

Week 1:

Views: Target 1,000+
CTR (Click-Through Rate): Target 8-12%
AVD (Average View Duration): Target 60%+ (9+ minutes)
Comments: Look for "This works!" testimonials
GitHub stars: Track repo engagement

Week 2-4:

Subscriber growth
Watch time accumulation
SEO ranking for "Claude Sonnet 4.5 resume"
External shares (Reddit, Twitter, LinkedIn)

Long-term:

Tutorial series continuation
Community building
Sponsor opportunities (Anthropic, job platforms)

🔄 ITERATION & IMPROVEMENT¶

Based on Analytics:¶

If AVD drops at specific timestamp → Re-edit that section
If CTR is low → Test new thumbnail
If comments ask for specific feature → Create follow-up video

Follow-up Video Ideas:¶

"Claude Sonnet 4.5 Cover Letter Agent"
"Interview Prep with Claude Sonnet 4.5"
"LinkedIn Profile Optimizer Agent"
"Salary Negotiation with AI"

✅ FINAL PRE-FLIGHT CHECK¶

Before you hit record:

[ ] Read this entire script 2-3 times
[ ] Customize experience databases with YOUR info
[ ] Find 3 real jobs to use as examples
[ ] Test the agent end-to-end at least once
[ ] Set up camera and lighting for on-camera segments
[ ] Configure OBS for screen recording
[ ] Clear desktop and disable notifications
[ ] Bookmark all necessary URLs
[ ] Prepare voiceover recording space (quiet, good mic)
[ ] Have water nearby for voiceover sessions
[ ] Energy level: HIGH (you're teaching something amazing!)

🚀 YOU'RE READY TO CREATE!¶

This script is your complete blueprint. Follow it section by section.

Remember:

Claude Sonnet 4.5 is your differentiator
Real-time demos build trust
Three professions show universal value
High energy keeps viewers engaged
Clear CTAs drive action

The message: People won't be replaced by AI. They'll be replaced by those who leverage AI.

You're teaching them how to leverage it.

Now go record something incredible. 🎬

Questions? Stuck on something? Check the backup folder or comment below!

📹 THE FULL SCRIPT - Episode 02¶

Resume Fine-Tuner with Claude Sonnet 4.5¶

🎬 VIDEO TITLES (SEO-Optimized Options)¶

⏱️ VIDEO STRUCTURE¶

🔥 HOOK (0:00-1:15)¶

Part 1: YOU ON CAMERA - Strong Open (5 seconds)¶

Part 2: Screen Share - Your Proof (45 seconds)¶

Part 3: You on Camera - The System (25 seconds)¶

📚 THE PROOF: 100+ RESUMES GENERATED (1:19-3:01)¶

📊 WHY CLAUDE SONNET 4.5? (3:01-4:01)¶

Part 1: Model Comparison (45 seconds)¶

Part 2: The Benchmarks That Matter (60 seconds)¶

Part 3: Live Claude Demo (45 seconds)¶

🎯 THE PROBLEM (4:01-4:46)¶

🤖 CLAUDE CODE AGENTS (4:46-5:20)¶

Part 1: What Are Agents? + System Diagram (90 seconds)¶

Part 2: Claude Desktop Token Tracking (30 seconds)¶

📁 GITHUB RESOURCES & SETUP (5:24-6:54)¶

Part 1: GitHub Repository & Documentation (60 seconds)¶

Part 3: Workspace Quick View (30 seconds)¶

🎨 CREATE THE AGENT (6:54-8:24)¶

Part 1: Agent Configuration File (90 seconds)¶

💼 DEMO 1: SOFTWARE ENGINEER (7:24-9:54)¶

Part 1: Create Experience Database (60 seconds)¶

Part 2: Find Real Job (30 seconds)¶

Part 3: Show the Result - LaTeX Resume Example (60 seconds)¶

🎨 OUTPUT FORMATS & OTHER PROFESSIONS (9:54-10:54)¶

Part 1: Output Formats - Your Choice (30 seconds)¶

Part 2: Templates for Other Professions - DIY (30 seconds)¶

🎨 ITERATIVE REFINEMENTS (10:54-11:24)¶

Part 1: Quick Adjustments - Conversational Editing (45 seconds)¶

📈 RESULTS & THE BIG PICTURE (11:24-12:13)¶

Part 1: Metrics That Matter (30 seconds)¶

Part 2: The Real Message (30 seconds)¶

🎬 CLOSING (12:13-12:58)¶

END SCREEN (20 seconds)¶

📊 PRODUCTION SUMMARY¶

What You'll Record:¶

✅ PRE-PRODUCTION CHECKLIST¶

Before Recording:¶

🎯 CRITICAL SUCCESS FACTORS¶

1. Show Real-Time Claude Demos (MOST IMPORTANT!)¶

2. High Energy On Camera¶

3. Use Real Jobs¶

4. Emphasize Claude Sonnet 4.5¶

5. Three Different Professions¶

6. The "AI Replacement" Message¶

🚨 COMMON MISTAKES TO AVOID¶

📈 POST-PRODUCTION & UPLOAD¶

YouTube Upload Checklist:¶

🎯 SUCCESS METRICS TO TRACK¶

🔄 ITERATION & IMPROVEMENT¶

Based on Analytics:¶

Follow-up Video Ideas:¶

✅ FINAL PRE-FLIGHT CHECK¶

🚀 YOU'RE READY TO CREATE!¶