Nerv - (Ai interviewer )
Welcome to NERV — The Smartest Mock Interviewer Powered by AI, NERV doesn’t just ask questions — it reads you. Using advanced facial analysis and your provided information, NERV generates tailored in
视频
描述
🚀 NERV — AI-Powered Mock Interviews
🔍 Problems :
✔ Static & Generic Interview Questions
✔ No Feedback on Emotions or Reactions
✔ No Personalization from Resume
✔ Lack of Real Conversation Flow
✔ Candidates Lack Confidence in High-Stress Scenarios
✔ Limited Accessibility to Quality Mock Interview Resources
🧠 How We Solved It:
✔ Real-time Facial Emotion Analysis using Hume AI to gauge emotional responses and provide behavioral feedback
✔ Dynamic Question Generation powered by GPT-4o + RAG, ensuring relevance and depth
✔ Resume Parsing from PDF to extract experience, skills, and education for personalized questions
✔ Voice-to-Text using Whisper AI for seamless conversation capture
✔ Text-to-Voice (TTS) to create a human-like interviewer experience
✔ Feedback Module that gives actionable insights on tone, confidence, and clarity
✔ Conversation History & Analytics to track growth over time
✔ Custom Interview Modes: Technical, HR, Behavioral, etc.
🌟 Impact:
✔ Adaptive Interviews that evolve based on your emotions and answers
✔ Hyper-Personalized Questions from your resume, real-time reactions, and job role
✔ Realistic Interview Flow for improved confidence and spontaneity
✔ Data-Driven Feedback to target your weak spots and boost strengths
✔ Practice Anytime, Anywhere — accessible mock interviews for everyone
💡 Future Scope:
Integration with LinkedIn & Job Portals for auto-importing resumes
Support for Multiple Languages & Accents
AI-powered Post-Interview Evaluation Report
Dashboard for Institutes/Training Centers to track candidate performance
Live Interview Mode with Mentor/HR Review
本次黑客松进展
🛠️ Progress During Hackathon During the hackathon, our team made significant strides in both development and implementation. Here's a breakdown of the key progress: - Implementation of RAG with GPT-4o We successfully integrated a Retrieval-Augmented Generation (RAG) system using OpenAI's GPT-4o model. This setup allowed us to: Retrieve relevant context from our custom dataset. Feed that context dynamically into GPT-4o for more accurate, contextual, and meaningful responses. Improve the reliability of answers, especially for domain-specific queries. This was a major milestone in enhancing the intelligence and relevance of our AI system. Ensured smooth interaction between the frontend and backend. Designed the system to be scalable and modular for future improvements. 📩 Feedback and Suggestions Received We received valuable feedback from mentors and evaluators, which helped us improve further: Feedback 1: Great job on using GPT-4o with RAG. Try optimizing your retrieval logic for faster results. We added indexing to the document store to improve retrieval speed. Feedback 2: The UI is clean, but try to give visual cues when the model is processing. Feedback 3: Ensure the system can handle unexpected or vague queries. - We added fallback messages and used GPT-4o’s capabilities to handle ambiguity more gracefully.
技术栈
融资状态
At this stage, we have not yet finalized our fundraising strategy. Our current focus remains on building a robust and impactful solution during the hackathon. We do, however, recognize the importance of funding for scaling and sustaining the project in the future. Once the initial version is stable and tested, we plan to explore various fundraising options, including grants, startup incubators, and strategic partnerships. Further planning and discussions around funding will be carried out in the upcoming phases.