I recently built AI Interview, an open-source AI voice interview platform.
A while ago, I came across a project called FoloUp on GitHub.
It had a great idea: AI voice interviews. A candidate gets a link, talks to an AI interviewer, and the recruiter gets a scorecard with feedback.
The project had more than 1,000 stars, so I wanted to try running it myself.
But when I looked at the setup, I saw that it depended on services like Clerk for authentication and Supabase for the database.
There is nothing wrong with that. Those tools are great for many products.
But I wanted something different.
I wanted an AI interview platform that I could run in my own deployment, with my own database, and without signing up for separate auth or database services.
So I built AI Interview.
It follows the same basic idea, but with a self-hosted setup:
- Better Auth instead of Clerk
- PostgreSQL instead of Supabase
- Vercel or Docker for deployment
- MIT license so anyone can use, change, and self-host it
The workflow is simple:
- Paste a job description or upload a PDF
- AI generates interview questions
- The candidate joins through a link
- The AI conducts the voice interview
- You get a scorecard with a full transcript
Candidates do not need to create an account or install anything. They can open the link and complete the interview in their browser.
AI Interview also includes custom interviewer personas, candidate tracking, tab-switch monitoring, and interview analytics.
No external auth service.
No Supabase dependency.
No seat-based pricing.
Just your server, your database, and your hiring workflow.
You still need to pay for the AI and voice APIs you use, like OpenAI and Retell AI. But the platform itself is self-hosted, so you are not locked into paid auth, database, or hiring software subscriptions.
Product page: https://ai-interview.brijeshkumaryadav.com
GitHub: https://github.com/brijeshmarch16/ai-interview