tuchg / source-cv

An LLM-powered personal resume builder tool that prioritizes privacy while effortlessly enabling data sharing and reuse among resumes.

SourceResume

Development

setup

Firstly, we should create a .env file in the root directory of project repository and write the following content into it.

OPENAI_API_BASE="https://api.openai.com/v1"
OPENAI_API_KEY="sk-XXXX"

run

pnpm install
pnpm run dev

Credits

• Resume-Schema

source-cv's Issues

Project Refactor

Refactoring code structure, improving reuse, optimizing performance, improving development experience, etc

• #10
• #8
• #9
• #5
• #12

Add client support with tauri

Add translate a resume version

base on LLM

add CI/CD support

Optimize the `preview-section` based on `JSON-RESUME`

enhance the user experience by adding features such as panning-tool , zooming functionality etc.

Remove user system

The user system is just something I use to pass my graduation project.

Add error and loading handling

Refactor the template capture function

To resolve the current issue of SVG exceeding the capacity of localstorage.

Refactor the database to use `SQLLite`

The current data access method is JSON, which is very primitive. SQLLite will be used to better support resume data

Risk:
Reduced compatibility, SQLLite requires the use of WASM technology on the browser side

Support for the resume data import

We will provide support for a resume data import feature, enabling users to easily import data from LinkedIn and PDF files.

• LinkedIn: data can be seamlessly obtained through the API
• PDF file: data will be extracted using LLM drivers specifically designed for this purpose.

Once the extraction process is complete, the data will be automatically organized and incorporated into the resume database, ensuring a streamlined and efficient workflow for users.

GithubBot

Add full i18n support

To refactor existing code to use `Next.js`

To automatically select data and generate resume based on LLM

Automatically select suitable resumes based on the job requirements specified by the user and refine the text style to generate resumes

Refactor the resume `preview-section` to split it into different template preview engine.

The current template preview is implemented based on WebContainers and JSON-RESUME, which allows integration with a vast template ecosystem offered by JSON-RESUME. Currently, there are over 400 themes available for users to choose from, greatly enriching their options.

However, these templates have varying implementation methods, making it difficult or even impossible for us to provide customization capabilities for end-users.

Therefore, we plan to develop our own templates specifically designed to offer customization capabilities. JSON-RESUME template will also continue to be supported.

The own template engine should be:

• Provide support for adjusting the page layout extensively to achieve a one-page PDF effect.
• Custom section such as rename,sort,font,theme color etc.
• Perhaps in the future, we will be able to support developers in creating templates that do not rely on the original repository for building, allowing users to directly import pre-packaged themes developed by developers.

Replace `material-react-table`

We should implement it ourselves table using Shadcn/UI to reduce the volume after packaging and ux

Features

• Paging List
• Assign
• Add
• Delete
• Edit
• Searching base on LLM