InterPrepAI

• 

  Ace Your Next Interview with AI-Powered Practice

Inspiration Interviews are nerve-wracking for many, especially fresh graduates, career switchers, and those without access to quality mentorship. We were inspired by the success of platforms like RoundOne.ai but wanted to go beyond audio-only mock interviews by integrating AI-powered voice + video simulations, personalized resume-based questioning, real-time scoring, and direct HR connect. We built InterPrepAI to create a fair, scalable, and engaging way for candidates to practice, improve, and even get discovered by recruiters.

What it does 1.Conducts realistic AI-powered interviews using voice and video. 2.Dynamically generates questions based on the candidate’s resume, job role, and skill level. 3.Provides real-time feedback and interview scoring based on voice, fluency, relevance, and confidence. 4.Candidates who cross an 80% score threshold are automatically shortlisted and shared with recruiters. 5.Enables HRs to schedule live audio/video interviews with shortlisted candidates, similar to RoundOne.ai but with video support.

How we built it 1.Frontend: React, TypeScript, TailwindCSS 2.Backend: Node.js + SupaBase 3.LLM-based Q&A: Gemini 4.Voice & Speech: Eleven Labs / Whisper / Google Speech-to-Text 5.Scoring: Sentiment analysis + keyword matching + confidence metrics 6.Video Calls: WebRTC (custom) 7.Resume Parsing: Python backend using spaCy + Pdf miner 8.Real-time Updates: WebSockets 9.Hosting: Vercel

Challenges we ran into 1.Voice & autoplay restrictions in browser widgets 2.Real-time synchronization of video, speech, and text analysis 3.Creating natural follow-up questions using LLMs 4.Scaling LLM usage under API token quotas 5.Getting video + voice scoring to work in harmony 6.Ensuring smooth fallback when network or mic/cam access fails

Accomplishments that we're proud of 1.Built a working voice+video interview AI agent from scratch 2.Developed real-time scoring and HR pipeline integration 3.Enabled resume-driven AI question generation 4.Created a full-stack platform deployable on the web 5.Simulated real interview experience with a clean UI/UX

What we learned 1.How to effectively combine LLMs, TTS/STT, and video streaming 2.Building LLM prompts that simulate real interviewers 3.Parsing resumes into structured skill maps 4.Managing API token quotas and optimizing prompt size 5.Designing a product that balances UX and AI feedback

What's next for InterPrepAI 1.Integrate live feedback coaching and personality analysis 2.Add multilingual support (Hindi, Spanish, etc.) 3.Enable HR dashboard analytics to view candidate pipeline 4.Expand to campus placements and bootcamp partnerships 5.Launch mobile app for both Android and iOS 6.Add peer-to-peer mock interviews powered by AI moderation 7.Partner with hiring firms to automate first-round interviews

Built With

• css
• elevenlabs
• gemini
• html5
• javascript
• react
• supabase
• typescript