CVAlchemix

Stop sending the same CV to every job. CVAlchemix scrapes LinkedIn postings and rewrites your CV with AI — tailored, compiled, and PDF-ready in seconds.

🎬 Demo

View demo GIF

✨ Features

• Scrapes LinkedIn job title, company, location, and full job description with Playwright.
• Reuses a persistent browser profile so you can keep a logged-in LinkedIn session.
• Prompts for a Gemini API key and a plain-text base CV, then stores them locally for later runs.
• Includes a built-in cvalchemix login command to open LinkedIn and persist an authenticated browser session.
• Rewrites the CV into a structured CVData schema using Google Gemini.
• Renders the final CV through a Jinja2 LaTeX template and compiles it to PDF with tectonic.
• Saves the intermediate .tex file alongside the PDF so you can inspect the generated LaTeX.
• Includes show-config and delete commands for inspection and cleanup.

🎯 Use Cases

• A job seeker can paste a LinkedIn posting and generate a tailored CV PDF that mirrors the role's keywords and structure.
• A candidate applying to several roles can reuse the same base CV and produce a separate PDF for each company and posting.
• A developer can automate a job-application workflow by combining the scraper, Gemini rewrite step, and PDF renderer in one CLI.
• Someone keeping a persistent LinkedIn session can avoid repeated logins when scraping job pages over time.
• A career coach or reviewer can inspect the generated .tex and final PDF to understand how the CV was reshaped for a role.

⚠️ Current Limitations

These are known limitations in the current version. They will be addressed in future releases.

• LinkedIn URL format: Only the following URL pattern is currently supported:

  https://www.linkedin.com/jobs/collections/recommended/?currentJobId={currentJobId}
  

  Other LinkedIn URL shapes such as /jobs/view/ and search-result pages are not yet supported.

• AI Provider: Currently only Google Gemini is supported as the AI provider.

• External renderer: PDF generation depends on a local tectonic binary being available on PATH.

• Scraping fragility: The scraper relies on LinkedIn CSS selectors, so LinkedIn DOM changes can break extraction.

• Template scope: Only one bundled LaTeX template, classic.tex.j2, is shipped right now.

• No built-in fit score: The repository contains structured analysis models, but there is no user-facing command that emits a standalone job-match score or analysis report yet.

🔩 Prerequisites

• Python >= 3.10
• A Google Gemini API key
• tectonic installed and available on your PATH

Playwright and its browsers are installed automatically as part of the Python dependencies — no manual step needed.

📦 Installation

Option 1 — pip

pip install cvalchemix

Option 2 — pipx (recommended for CLI tools)

pipx install cvalchemix

Option 3 — One-command install (macOS/Linux)

curl -fsSL https://raw.githubusercontent.com/kayesFerdous/CVAlchemix/main/install.sh | bash

If you don't have curl:

wget -qO- https://raw.githubusercontent.com/kayesFerdous/CVAlchemix/main/install.sh | bash

Option 4 — Windows (PowerShell)

irm https://raw.githubusercontent.com/kayesFerdous/CVAlchemix/main/install.ps1 | iex

The installer scripts prefer pipx and fall back to pip --user when needed. You can override the install source with the CVALCHEMIX_INSTALL_TARGET environment variable.

✅ Verify Installation

cvalchemix --help

🚀 Quick Start

1. Configure the app with your Gemini key and a plain-text base CV.

cvalchemix configure

2. Optional but recommended — open a persistent LinkedIn session once so the scraper can reuse it.

cvalchemix login

3. Generate a tailored CV from a LinkedIn job URL.

cvalchemix generate "https://www.linkedin.com/jobs/collections/recommended/?currentJobId=1234567890" -o ./output

The CLI writes the final PDF to ./output/cv/<company>_<timestamp>/cv.pdf and saves the intermediate LaTeX source as cv.tex in the same directory.

📖 Usage

Command	What it does	Example
cvalchemix configure	Prompts for a Gemini API key and the path to your base CV text file, then saves them in the local config file.	cvalchemix configure
cvalchemix login	Opens LinkedIn in a persistent Playwright browser profile and stores login readiness for later generate runs.	cvalchemix login
cvalchemix show-config	Displays the saved configuration and masks the stored API key.	cvalchemix show-config
cvalchemix generate <job-url>	Scrapes a LinkedIn job post, rewrites your CV with Gemini, and renders a PDF. Use -o or --output to set the destination directory.	cvalchemix generate "https://www.linkedin.com/jobs/collections/recommended/?currentJobId=1234567890" -o ./output
cvalchemix delete	Removes local CVAlchemix data and uninstalls the package by default. Use --data-only to keep the CLI installed, and -y to skip confirmation.	cvalchemix delete -y

⚙️ Configuration

CVAlchemix uses two layers of configuration:

• CLI config — run cvalchemix configure to save your gemini_api_key and base_cv_path into a local JSON config file under your platform user config directory.
• LinkedIn session — run cvalchemix login once to open LinkedIn and save linkedin_login_configured for generate preflight checks.
• A .env file in the working directory is also read if present.

Environment Variables

Variable	Default	Purpose
GOOGLE_API_KEY	empty	Gemini API key used by the settings layer.
DEFAULT_MODEL	gemini-2.5-flash-lite	Gemini model name used by the LLM wrapper.
PROFILE_DIR	managed app profile directory	Playwright browser profile location for persistent LinkedIn sessions.
OUTPUT_DIR	./output	Default output directory used by the settings layer.
BROWSER_HEADLESS	false	Launch Playwright in headless mode when set to true.
SCRAPE_TIMEOUT_MS	30000	Timeout in ms while waiting for LinkedIn page content to load.
MAX_RETRIES	3	Number of retries defined by the configuration layer.
LOG_LEVEL	INFO	Root logging level.

A .env.example file with starter values for all of the above is included in the repository.

🗺️ Roadmap

• OpenAI / Anthropic provider support
• Additional CV templates
• Job-fit score command
• Broader LinkedIn URL support

🛠️ Local Development

git clone https://github.com/kayesFerdous/CVAlchemix.git
cd CVAlchemix
./install.sh        # macOS/Linux
.\install.ps1       # Windows PowerShell

For a development install with editable mode:

pip install -e ".[dev]"

🤝 Contributing

1. Fork the repository and create a focused branch for your change.
2. Make the smallest practical change and keep the existing CLI behaviour intact unless the change explicitly requires otherwise.
3. Verify the project still starts, configure the CLI, and run a sample generate flow if your change touches the pipeline.
4. Open a pull request with a clear description of the change, the motivation, and any manual verification you performed.

📄 License

This project is licensed under the MIT License. See LICENSE for details.