🚀 MockMate — AI-Powered Placement & Multi-Panel Interview Simulation Platform

💡 Why We Built It (The Problem)

Most engineering students and job seekers don't fail interviews due to a lack of technical knowledge—they fail because of interview anxiety, unstructured answers, behavioral blindspots, and lack of real panel exposure:

1. Static Mock Tools Aren't Real: Text-based chat prompts or basic chatbots fail to replicate the pressure of facing senior interviewers firing spontaneous follow-ups.
2. Behavioral & Communication Flaws: Filler words ("umm", "like"), casual slang, and non-STAR method answers disqualify candidates before their coding scores are even tallied.
3. Generic Feedback & Language Barriers: Traditional tools give boilerplate advice like "Improve communication". Furthermore, non-native English speakers often struggle to grasp complex feedback when it isn't explained conversationally.

We built MockMate to bridge the gap between campus preparation and corporate placement with a hyper-realistic, multi-round AI simulation pipeline.

⚡ What This Build Is About

MockMate is an end-to-end interview intelligence platform featuring:

1. 👑 Game of Fours (4-Avatar Executive Panel):

   • A multi-interviewer panel interview simulating executive hiring rounds for students and experienced candidates alike.
   • Powered by HeyGen LiveAvatar streaming video with realistic lip-syncing and facial expressions across 4 distinct personas:
     • 💼 June (HR Manager): Evaluates culture fit, teamwork, and behavioral alignment.
     • ⚡ Bryan (Tech Lead): Probes code architecture, databases, and scalability trade-offs.
     • 🎯 Graham (Product Manager): Challenges product sense, user experience, and prioritization.
     • 👑 Alessandra (Hiring Manager): Assesses leadership, ownership under pressure, and business ROI.
   • Features intelligent keyword routing (routes your answer dynamically to the relevant expert) and a 5-second silence thinking buffer so candidates can pause and think without being cut off.

2. 🎙️ 1-on-1 Deep-Dive Review Panel (Powered by Agora Conversational AI):

   • Post-interview mentor debrief powered by Agora Conversational AI Agent (v2) with ultra-low latency, bidirectional WebRTC audio and native Voice Activity Detection (VAD).
   • Bilingual Mentorship (English & Hindi): Mentors can explain technical shortcomings and feedback in both English and Hindi, ensuring candidates completely understand their loopholes.
   • Real-Time & Interruptible: Candidates can speak naturally and interrupt the mentor mid-speech to ask clarifying questions.
   • Evidence-Based Feedback: Mentors quote specific lines from the candidate's actual interview transcript to explain what was missing and how to answer better.

3. 🛡️ Proctored Technical Round for Students:

   • Smart ATS Resume Parser: Extracts PDF resumes, scores keyword density, and personalizes question difficulty based on candidate experience.
   • Visual Anti-Cheating Proctoring: Live webcam monitoring checking for tab switching, gaze distraction, multiple persons, and unauthorized mobile devices via a 3-heart life system.
   • Behavioral Tone & Slang Auditing: Real-time auditing for filler words, casual slang, and unprofessional language.

4. 💼 June HR 1-on-1 Avatar Round:

   • Dedicated conversational HR screening with June HR featuring automated speech turn-taking and dynamic STAR-method question evaluation.

5. 📊 Personalized Candidate Performance Dashboard & GuideBot:

   • Resume Health Metric: Circular score with missing ATS keywords and a 1-click "Fix with AI" feature.
   • 3D Floating Skill Bubbles: Physics-animated skill tags parsed from the resume.
   • Performance vs. Pro Benchmarks: Comparative graphs evaluating Technical, Communication, and Confidence scores against industry standards.
   • Answer Blueprint Comparison: Side-by-side breakdown comparing the candidate's raw answer against an ideal STAR-method response.
   • Interactive GuideBot ("CuteBot"): Floating companion offering pre-interview DOs & DON'Ts checklists and hover tips.
   • Downloadable PDF Performance Cards: Comprehensive diagnostic certificates generated client-side via @react-pdf/renderer.

🌍 How It Can Be Useful For Others

• University Students & Placement Cells: Practice under real proctored conditions with instant ATS scoring and behavioral correction before campus placement drives.
• Experienced Job Seekers: Experience realistic executive panel dynamics (Game of Fours) with cross-functional questions from Tech, Product, HR, and Hiring Managers.
• Diverse Linguistic Backgrounds: Candidates can debrief their performance in Hindi or English, removing language intimidation and ensuring actionable growth.
• Bootcamps & Recruiters: Automated candidate screening reports with objective scoring, vision proctoring integrity logs, and downloadable diagnostic PDFs.

🛠️ Tech Stack

• Frontend: Next.js 14 (App Router), React 18, TypeScript 5, Tailwind CSS, Framer Motion
• Voice Intelligence: Agora Conversational AI Agent (v2 REST API & RTC WebRTC SDK)
• Video Avatars: HeyGen LiveAvatar Web SDK (Real-time video streaming & lip-syncing)
• LLM Engine: Google Gemini (@google/genai), Groq Llama-3 (llama-3.1-8b-instant), OpenAI
• Auth & Database: Google OAuth (@react-oauth/google), Supabase (PostgreSQL)
• Document Generation: @react-pdf/renderer

🔗 Relevant Links

• GitHub Repository: https://github.com/kavyabhardwaj2004/mockmate