🎯 AceIt: Real-Time Multimodal AI Interview Coach

Team Name: Mitchell

Team Members:

1. Jason Seh
2. Mitchell Eickhoff
3. Konrad Gozon
4. Rocky Shao

---

🚀 Project Overview

AceIt is a high-performance, AI-driven mock interview platform that provides real-time multimodal analysis. It combines computer vision (MediaPipe) and audio analysis with large language models to provide instantaneous feedback on candidate performance.

✨ Key Features

• Real-Time Biometrics: Tracks gaze stability, confidence levels, and fidgeting in real-time.
• "Response-First" Architecture: Sub-500ms AI response latency using Gemini Flash 3.0 and ElevenLabs.
• Specialized Tracks: Behavioral, System Design, and Technical (Coding) interview modes.
• Integrated Python IDE: A browser-based Python compiler (Pyodide) for real-time technical assessments.
• AI Coaching Reports: Detailed feedback with comparative "Candidate Action" vs. "Ideal Strategy" analysis.

🛠️ Performance Tech Stack

• Frontend: Next.js 15, TypeScript, Zustand, WebRTC, MediaPipe, TalkingHead (3D Avatar).
• Backend: Python aiohttp, aiortc, PyTorch.
• AI Intelligence: Google Gemini (Flash 3.0 & Pro 2.5), OpenAI Whisper, ElevenLabs Flash.
• Infrastructure: Supabase (PostgreSQL), Clerk.

📖 Documentation

Detailed documentation for developers and technical stakeholders:

• System Architecture
• Project Master Document

🏗️ Getting Started

Prerequisites

• Node.js 18+
• Python 3.10+
• Supabase Account
• Clerk Account

Installation

1. Clone the repository

   git clone [repository-url]
   cd HackAI2026

2. Frontend Setup

   npm install
   npm run dev

3. Backend Setup

   # Create and activate virtual environment
   python -m venv .venv
   source .venv/bin/activate # Windows: .venv\Scripts\activate
   
   # Install dependencies
   pip install -r requirements.txt
   
   # Start the backend server
   python backend/server.py

---

Built for the HackAI 2026 Hackathon.