Turn Your Resume/LinkedIn into a Personal Website

self.so is a 100% free and open-source AI site builder that takes your existing resume (as a PDF) or your LinkedIn profile and automatically generates a clean, professional-looking personal website.

No complex drag-and-drop builders, no picking templates for hours. You feed it your info, and it spits out a site. This makes it ideal for job seekers and freelancers looking to present their qualifications online.

Features

PDF and LinkedIn Import – Upload your existing resume or connect your LinkedIn profile to auto-generate your site
100% Free & Open Source – No hidden fees or premium tiers
AI-Powered – Uses Llama 3.3 (with plans to upgrade to Qwen 2.5 72B) to extract and organize your professional information
Modern Tech Stack – Built with Next.js app router, Together.ai for LLM functionality, and Vercel’s AI SDK
Secure Authentication – Integrated with Clerk for user management
Cloud Storage – S3 integration for PDF storage
Data Management – Upstash Redis database for fast performance
Analytics – Helicone integration for usage monitoring
Custom Domains – Host on your personal domain

Use Cases

Job Seekers – Create a professional online presence quickly during a job search without paying for website hosting
Freelancers – Generate a portfolio site to share with potential clients
Students – Build a professional online presence before entering the job market
Career Changers – Showcase your transferable skills in a visual format
Technical Professionals – The site automatically organizes your technical skills in a clean, scannable format

How To Use It

Create an account – Sign up at self.so using the Clerk authentication system
Upload your resume – Submit your PDF resume, which gets stored in S3 and scanned for safety using Llama Guard
Wait for processing – The system sends your PDF to an LLM (currently Llama 3.3) to extract key information in structured JSON format
Review and publish – Check your auto-generated site on a dynamic route and publish when satisfied

The whole process takes just minutes, with the AI doing most of the heavy lifting to organize your information in a professional layout.

Cloning & Running Locally (For the Devs)

This is where it gets interesting if you like to tinker or customize. Since it’s open source:

1. Fork/Clone: Grab the code from their GitHub repository.

2. Get API Keys: You’ll need accounts and API keys for:

Together AI: For access to the LLM (Llama 3.3 or future models).
Upstash: For the Redis database (handles user data, site info).
AWS: For an S3 bucket to store the uploaded PDFs.
Clerk: For handling user authentication in your local setup.

3. Configure Environment: Create a .env file (you can copy .example.env) and inser your API keys.

S3_UPLOAD_REGION=
S3_UPLOAD_KEY=
S3_UPLOAD_SECRET=
S3_UPLOAD_BUCKET=

UPSTASH_REDIS_REST_URL=
UPSTASH_REDIS_REST_TOKEN=

TOGETHER_API_KEY=

NEXT_PUBLIC_CLERK_PUBLISHABLE_KEY=
CLERK_SECRET_KEY=

4. Install & Run:

Run pnpm install to grab all the dependencies.
Run pnpm run dev to start the Next.js development server.

Pros

Genuinely Free: No cost to use the main service.
Open Source: Huge plus for transparency, customization, and learning.
Fast: Creates a site almost instantly after PDF processing.
Simple Interface: Very easy to use, minimal learning curve.
Modern Stack: Uses current, popular technologies.
Good Starting Point: Provides a functional base site quickly.

Cons

Basic Output: The generated site is clean but quite simple in terms of design and features (as of now).
Limited Customization (UI): Currently, you can’t edit sections or change themes directly through the web interface – though this is on their roadmap. Forking the repo is the current path for deep customization.
Requires External Accounts for Self-Hosting: If you run it yourself, you need accounts for Together AI, AWS S3, and Upstash, which might have their own free/paid tiers.
Parsing Accuracy: Relies on the LLM’s ability to understand your resume format. Highly unconventional resumes might yield imperfect results.
Future Features Still Pending: Things like editing links/sections, themes, and Helicone observability are planned but not yet implemented.

Future Roadmap Items

Switching the LLM from Llama 3.3 to Qwen 2.5 72B (potentially improving extraction).
Adding Helicone for better observability (useful for debugging AI calls).
Implementing error logging.
Allowing users to directly access their preview page.
Adding the ability to edit links and other sections directly on the site.
Introducing themes (starting with a “ghibli” theme).
Automatically deleting the previous resume from S3 when a new one is uploaded.

Related Resources

Together.ai: The LLM provider: https://www.together.ai/
Vercel AI SDK: The framework used for LLM interaction: https://sdk.vercel.ai/
Upstash: Redis database provider: https://upstash.com/