InterviewMate – AI Interview Platform
Me and my friend built InterviewMate during a hackathon after noticing how messy and inconsistent interviews can be for both candidates and interviewers.
The idea was simple. Help interviewers generate better questions and help candidates practice in a more realistic way.
We built a platform where:
- Interviewers get AI-generated questions based on the role
- Candidates can take mock interviews with AI
- The system evaluates answers and gives feedback
One important decision we made was around real-time communication. Instead of building everything from scratch using WebRTC, we used Stream API. Since it was a hackathon, we didn’t have time to deal with low-level infrastructure, so this helped us move faster and focus more on the product and AI side.
On the backend, we used a microservices-based approach. We handled real-time flows using WebSockets and built a system that manages interview sessions, generates questions using AI, and evaluates responses. We also designed the database to properly store sessions, reports, and user data.
We also spent a good amount of time tuning prompts so the AI actually gives useful questions and feedback instead of generic answers.
Tech stack
- Next.js, TypeScript, Tailwind CSS
- Node.js, Express, Socket.io
- PostgreSQL, Prisma
- OpenAI API
- Stream API (for real-time video calls)
This project ended up getting selected in the top 25 out of 1000+ projects for the T-Hub funding round, which was honestly a great validation for us.