An ultra-fast, intelligent mock interview and voice assessment platform designed to simulate authentic engineering rounds. It conducts deep resume parsing, hosts interactive voice interviews with real-time speech-to-text, provides hints upon hesitation, and compiles comprehensive hiring committee evaluation dossiers — powered by Groq LPU high-speed inference.
- Continuous Duplex Experience: The interviewer and candidate communicate naturally as if in a live phone call.
- Automatic Microphone Activation: The candidate's microphone automatically activates upon entering the call and immediately turns back on when the AI finishes speaking.
- Natural Silence & Hesitation Cadence: A calibrated 2.4s pause threshold lets candidates collect their thoughts without premature interruptions, with gentle hints provided if stuck for >14s.
-
Real-Time Speech Normalization & Domain Dictionary: Resolves common phonetic mistranscriptions for tech terms:
-
rock/grok$\rightarrow$ Groq -
rag us/ragas$\rightarrow$ RAGAS,rag$\rightarrow$ RAG -
quadrant$\rightarrow$ Qdrant -
pine cone$\rightarrow$ Pinecone -
lang chain$\rightarrow$ LangChain,lang graph$\rightarrow$ LangGraph -
fast api$\rightarrow$ FastAPI,next js$\rightarrow$ Next.js,react js$\rightarrow$ React -
super base$\rightarrow$ Supabase,post gres$\rightarrow$ PostgreSQL -
k8s$\rightarrow$ Kubernetes,ast$\rightarrow$ AST,dsa$\rightarrow$ DSA, etc.
-
- Animated Interviewer Avatar: Lifelike person avatar with natural eye-blinking and subtle mouth-speaking animations during spoken questions.
- Granular Parsing: AST and structural resume extraction of verified skills, target roles, project architectures, past employment achievements, and honors.
- Hiring Committee Flaw Screener: Pinpoints resume anti-patterns (e.g. missing quantifiable impact metrics, passive phrasing, dense skill lists) with actionable before-and-after elevation tips.
- Technical Round: In-depth probing on DSA, system architecture scalability, runtime complexity, and tech stack internals.
- Managerial Round: Evaluates technical trade-offs, engineering leadership, project ownership, and resolving cross-functional friction.
- HR & Cultural Round: Behavioral STAR methodology, company values alignment, and career vision.
- Group Discussion (GD) Round: Continuous 5-minute uninterrupted spoken delivery on contemporary topics covering Introduction, For, Against, and Conclusion.
- Dynamic 4-Criteria Scorecards: Granular performance breakdown tailored to the specific round type.
- Executive Summary Memo: Lead hiring committee review memo outlining overall hire recommendation, strengths, and areas for improvement.
- Question-by-Question Benchmark: Full audit of candidate transcript responses scored against model benchmark answers.
- Export to Dossier / TXT: Downloadable hiring dossiers for offline review.
- Floating Capsule Navbar: Soft rounded pill navigation bar with active item pill highlights (
#f2f3f3light background,#282d35dark background). - Dark & Light Mode Themes: Tailored HSL surfaces with seamless transitions and local storage persistence.
- Interactive Illustration Hero: High-resolution hero banner with clean non-boxed PNG illustration and quick launch triggers.
| Layer | Technology |
|---|---|
| Frontend Framework | React 19.2 + TypeScript 6 |
| Bundler & Dev Server | Vite 8 |
| Styling & Design System | Tailwind CSS v4 + Material 3 Tokens |
| Inference Engine | Groq LPU (llama-3.3-70b-versatile) |
| Voice & Speech | Web Speech API (webkitSpeechRecognition + speechSynthesis) |
| Database & Cache | Supabase JS Client + LocalStorage fallback |
| Icons | Lucide React |
| PDF Extraction | Client-side PDF.js worker |
- Node.js >= 18
- npm >= 9
- A free Groq API Key (Get one at console.groq.com)
# Clone the repository
git clone https://github.com/LikhithSP/AI-Interview-Preparation-Assistant.git
cd AI-Interview-Preparation-Assistant
# Install dependencies
npm install
# Start Vite development server
npm run dev| Command | Description |
|---|---|
npm run dev |
Starts local dev server at http://localhost:5173 |
npm run build |
Runs TypeScript compilation (tsc -b) and Vite production bundle |
npm run lint |
Runs linter |
npm run preview |
Previews production build locally |
The app functions completely offline with localStorage and requires only your Groq API key:
- Enter your key in the in-app Settings modal (top right gear icon), OR
- Create a
.envfile in the project root:
VITE_GROQ_API_KEY=gsk_your_groq_api_key_hereIf you wish to synchronize records to cloud Supabase rather than local storage:
VITE_SUPABASE_URL=https://your-project.supabase.co
VITE_SUPABASE_ANON_KEY=your_supabase_anon_keysrc/
├── assets/
│ └── interview-illustration.png # Hero banner illustration
├── components/
│ ├── App.tsx # Top bar, theme switch, view router
│ ├── Dashboard.tsx # Capsule navbar, hero, resume critique, dossier history
│ ├── MockInterview.tsx # Full-duplex voice call, animated avatar, STT normalizer
│ ├── InterviewSetupModal.tsx # 3-step round configuration & resume ingestion
│ ├── Evaluation.tsx # Hiring committee scorecard, feedback memo, model answers
│ └── SettingsModal.tsx # API key & preferences modal
├── services/
│ ├── groq.ts # Groq LPU prompts (resume parser, conversational turn, reports)
│ └── supabase.ts # Hybrid Supabase / LocalStorage database provider
├── utils/
│ └── pdfParser.ts # Dynamic PDF text parser
├── index.css # Design tokens, Google Sans typography, avatar keyframes
└── main.tsx # React entry point
- Chrome / Microsoft Edge (Recommended): Optimal support for full-duplex SpeechRecognition and Synthesis.
- Safari / Firefox: Fallback manual typing and native audio synthesis supported.
Distributed under the MIT License.