AI CV Booster

A .NET 8 WPF desktop application that uses Google Gemini AI to analyze, score, and rewrite CVs for clarity, impact, and ATS optimization — without fabricating any information.

---

Key Features

• PDF & DOCX Parsing — Upload CVs via drag-and-drop or file picker; text extracted using PdfPig and OpenXml
• AI-Powered Rewrite — Google Gemini analyzes and rewrites your CV with stronger action verbs, better structure, and ATS-friendly keywords
• CV Scoring — Receives a 0–100 score with detailed weakness analysis and keyword suggestions
• Job Description Targeting — Paste a job posting to tailor the rewrite with role-specific keywords
• Aggressive Mode — Toggle for a bolder, high-agency writing tone
• Side-by-Side Comparison — View original and improved CV text in parallel
• Export — Copy to clipboard or save as .txt
• Fully Async — All AI and parsing work runs off the UI thread; the interface never freezes
• Cancelable — Cancel mid-analysis from the loading screen
• Truth-Preserving — The AI is explicitly instructed not to fabricate credentials, employers, dates, or qualifications

---

Tech Stack

Layer	Technology
Language	C# (latest) / .NET 8.0
UI Framework	WPF (net8.0-windows)
Architecture	MVVM via CommunityToolkit.Mvvm 8.3.2
DI & Config	Microsoft.Extensions.Hosting / DependencyInjection / Configuration 8.x
UI Theme	MaterialDesignThemes 5.1.0 + MahApps.Metro 2.4.10 (Dark theme, DeepPurple/Lime)
PDF Parsing	UglyToad.PdfPig 1.7.0
DOCX Parsing	DocumentFormat.OpenXml 3.1.0
AI Provider	Google Gemini API (REST via HttpClient)
HTTP	Microsoft.Extensions.Http with 90s timeout + exponential backoff retries

---

Prerequisites

Requirement	Details
OS	Windows 10 or later (WPF — Windows only)
.NET SDK	.NET 8.0 SDK
Gemini API Key	Obtain from Google AI Studio

---

Installation & Setup

1. Clone the repository

git clone <repository-url>
cd new_cv

2. Configure the Gemini API key

Copy the sample config and add your API key:

cp src/AiCvBooster/appsettings.sample.json src/AiCvBooster/appsettings.json

Edit src/AiCvBooster/appsettings.json:

{
  "Gemini": {
    "ApiKey": "YOUR_GEMINI_API_KEY_HERE",
    "Model": "gemini-2.0-flash",
    "BaseUrl": "https://generativelanguage.googleapis.com/v1beta"
  }
}

Note: appsettings.json is listed in .gitignore and will not be committed. Never share your API key.

3. Restore and build

dotnet restore
dotnet build src/AiCvBooster/AiCvBooster.csproj -c Debug

4. Run

dotnet run --project src/AiCvBooster

Or run the compiled binary directly:

src\AiCvBooster\bin\Debug\net8.0-windows\AiCvBooster.exe

---

Usage Guide

1. Upload — Drag a .pdf or .docx file onto the drop zone, or click to browse
2. Target (optional) — Paste a job description into the targeting field to tailor keyword optimization
3. Aggressive Mode (optional) — Enable for a bolder, more assertive rewrite tone
4. Boost — Click Boost My CV to start analysis
5. Review — Examine the score (0–100), weaknesses, suggested keywords, and the rewritten CV side-by-side
6. Export — Copy the improved CV to clipboard or save as a .txt file

Configuration Options

Setting	Description	Example Values
Gemini:ApiKey	Your Google Gemini API key	AIzaSy...
Gemini:Model	Gemini model to use	gemini-2.0-flash, gemini-1.5-pro, gemini-1.5-flash
Gemini:BaseUrl	API endpoint	https://generativelanguage.googleapis.com/v1beta

---

Preview

Architecture & Technical Details

Project Structure

AiCvBooster.sln
src/AiCvBooster/
├── App.xaml(.cs)           # DI host bootstrap, global exception handlers
├── appsettings.json        # Runtime config (not committed)
├── appsettings.sample.json # Template for setup
├── Models/                 # CvDocument, AnalysisRequest, CvAnalysisResult, AppSettings
├── Services/               # CvParserService, GeminiClient, GeminiCvService, PromptBuilder, DialogService
├── ViewModels/             # MainViewModel, UploadViewModel, LoadingViewModel, ResultViewModel
├── Views/                  # MainWindow + Upload/Loading/Result XAML views
├── Resources/              # Colors, Typography, Controls (XAML resource dictionaries)
└── Converters/             # BoolToVisibilityConverter, ScoreToBrushConverter

Data Flow

PDF/DOCX File
    → CvParserService (text extraction)
    → CvDocument { RawText, Source }
    → PromptBuilder (constructs AI prompt with job desc + mode)
    → GeminiClient (HTTP POST to Gemini API, retries on 429/5xx)
    → GeminiCvService (JSON parse response)
    → CvAnalysisResult { Score, Weaknesses, Keywords, ImprovedText }
    → ResultViewModel (displayed in UI)

Key Design Decisions

• MVVM + DI — ViewModels are injected via Microsoft.Extensions.DependencyInjection; Views bind through XAML DataContext
• Service Abstraction — AI logic sits behind IAiCvService, making the Gemini provider swappable
• Resilient HTTP — GeminiClient implements exponential backoff (1s → 3s → 6s) for transient failures (429, 5xx, timeouts)
• Structured Errors — All AI failures are wrapped in AiServiceException with an AiFailureKind enum for clean UI error messaging
• State Machine — MainViewModel manages view transitions: Upload → Loading → Result (or back to Upload on error/cancel)

---

Troubleshooting

Common Issues and Fixes

Problem	Cause	Fix
App crashes on startup with no UI	Missing or malformed appsettings.json	Copy appsettings.sample.json to appsettings.json and fill in your API key. Check startup-error.log in the app directory for details
"Authentication" error on analysis	Invalid or expired Gemini API key	Verify your API key at Google AI Studio and update appsettings.json
"RateLimited" error	Too many requests to Gemini API	Wait a minute and retry. The app already retries with backoff, so this means the quota is genuinely exhausted
"Network" error	No internet or firewall blocking generativelanguage.googleapis.com	Check connectivity. Ensure your firewall/proxy allows HTTPS to generativelanguage.googleapis.com
"InvalidResponse" error	Gemini returned unexpected JSON	Try a different model in appsettings.json (e.g., gemini-2.0-flash). Some preview models have inconsistent output
PDF text is empty or garbled	Scanned/image-based PDF without embedded text	Use a DOCX file instead, or run the PDF through OCR software first
Build fails: "net8.0-windows" not found	.NET 8 SDK not installed or wrong OS	Install the .NET 8 SDK. This app requires Windows
UI elements look broken	Missing NuGet packages	Run dotnet restore before building
90-second timeout	Very large CV or slow API response	Split the CV into fewer pages, or try a faster model like gemini-2.0-flash

Diagnostic Files

• startup-error.log — Written to the app directory if the application crashes during startup. Contains the full exception stack trace.

---

Contributing

1. Fork the repository
2. Create a feature branch: git checkout -b feature/your-feature
3. Make your changes, ensuring they follow the existing MVVM architecture
4. Test locally with dotnet build and dotnet run --project src/AiCvBooster
5. Submit a pull request with a clear description of what changed and why

Guidelines

• Follow the existing service abstraction pattern — new AI providers should implement IAiCvService
• Keep ViewModels free of direct UI dependencies
• Do not commit appsettings.json or any API keys

---

License

See LICENSE for details.