Intix

Intix is an AI interview coach that simulates real interview conditions and gives measurable feedback on content quality, speaking delivery, and non-verbal communication.

I built this from a personal problem: when I prepared for interviews, most mock tools felt shallow. They either generated generic questions or gave feedback without enough detail to improve. Intix was built to close that gap with one practical, end-to-end workflow.

Core Capabilities

• Resume-based interview planning (optional)
• Role and difficulty-aware question generation
• Speech analysis (clarity, pace, filler words, delivery patterns)
• Visual analysis (eye contact, posture, expressions, gestures)
• Session-level scoring and actionable recommendations
• Session history tracking for progress over time

Why This Project Stands Out

• Built from a real user pain point, not just a feature demo
• Combines answer quality, speech, and visual signals in one system
• Supports both resume and no-resume interview modes
• Includes engineering practices expected in production-style projects

Security and Reliability Implemented

This project includes concrete backend security controls, not only feature logic.

1. API Rate Limiting

• In-memory rate limiting protects endpoints from abuse and noisy traffic.
• Returns 429 Too Many Requests with Retry-After header.
• Adds X-RateLimit-Remaining response header for transparency.

2. Strict Input Validation and Sanitization

• Validates role and difficulty against allowlists.
• Sanitizes text input and enforces max lengths.
• Validates audio/base64 payload structure and estimated size.
• Sanitizes uploaded filenames to prevent path or unsafe-character issues.

3. Safe File Upload Handling

• Restricts resume uploads to allowed MIME types (PDF, TXT).
• Enforces maximum file size limits via configurable settings.
• Rejects malformed or empty uploads early.

4. Controlled CORS Policy

• Uses explicit origin allowlist (environment-driven), not unrestricted wildcard origins.
• Supports required browser preflight behavior while keeping origin scope controlled.

5. Error Safety (No Sensitive Leakage)

• Centralized error handling and sanitization.
• Sensitive internal details can be hidden in production responses.
• Structured error responses with consistent shape and status handling.

6. Request Traceability and Observability

• Generates request IDs for traceability (X-Request-ID).
• Tracks per-request response time (X-Response-Time).
• Uses rotating log files to preserve diagnostics while controlling log growth.

7. Defensive Service Initialization

• Optional heavy dependencies (vision stack) are loaded with safe fallback.
• API remains operational with graceful degradation when optional services are unavailable.

Architecture Overview

• Frontend: React + Vite + Tailwind CSS
• Backend: FastAPI + Pydantic
• Analysis Pipeline: AI question/evaluation + speech analytics + computer vision + confidence scoring

Local Setup

Backend

cd backend
python -m venv venv
venv\Scripts\activate
pip install -r requirements.txt
python main.py

Frontend

cd frontend
npm install
npm run dev

Project Intent

Intix is designed as a practical, recruiter-relevant engineering project: strong product thinking, measurable AI feedback, and visible security and reliability practices that reflect real software development standards.