FortiTwin is a modern, full-stack AI interview platform designed to revolutionize the recruitment process. It acts as a "digital twin" of a professional interviewer, conducting technical and behavioral assessments with real-time feedback.
Unlike traditional forms, FortiTwin utilizes RAG (Retrieval Augmented Generation) to analyze candidate resumes and generate context-aware questions. It also features an Emotion Engine to analyze vocal sentiment (confidence, nervousness) during the interview.
- 🧠 AI Neural Core: Powered by Gemini/OpenAI to conduct dynamic, human-like conversations.
- 📄 Smart Resume Analysis: Asynchronously parses resumes to tailor interview questions to the candidate's specific skills.
- 🗣️ Voice & Emotion Analysis: Integrated with Hume AI to detect vocal cues like confidence and nervousness in real-time.
- 💻 Technical Sandbox: Live coding environment for assessing programming skills (supports Python/JS).
- 📊 Comprehensive Reports: Generates detailed scorecards covering technical accuracy, communication skills, and behavioral traits.
- ⚡ Async Architecture: Uses Redis and ARQ to handle heavy workloads (PDF parsing, vector embedding) without blocking the UI.
The project follows a monorepo structure:
| Service | Path | Tech Stack | Description |
|---|---|---|---|
| Web App | /apps/web |
Next.js 14, TypeScript, Prisma, Tailwind | The user-facing frontend for Candidates and HR. |
| AI Engine | /apps/ai-engine |
FastAPI, Python, LangChain, Qdrant | The brain handling LLMs, RAG, and Voice processing. |
| Database | - | MongoDB | Primary data store for users and assessments. |
| Queue | - | Redis + ARQ | Background job processing for resume ingestion. |
Ensure you have the following installed:
- Node.js (v18+)
- Python (3.10+)
- Docker (optional, for running Redis/Mongo easily)
- MongoDB (or a cloud Atlas URI)
- Redis (required for the AI Engine queue)
git clone [https://github.com/yourusername/fortitwin.git](https://github.com/yourusername/fortitwin.git)
cd fortitwin