1. Getting Started

1.1. System Requirements

Before deploying MAC, ensure your system meets the following requirements:

Minimum Requirements
Component	Requirement
Operating System	Windows 10/11, Ubuntu 20.04+, or macOS 12+
Docker Desktop	Version 4.0+ with Docker Compose v2
RAM	8 GB minimum (16 GB recommended for LLM inference)
Storage	20 GB free disk space (50 GB with models)
GPU (optional)	NVIDIA GPU with 8+ GB VRAM (e.g., RTX 3060 12 GB)
Network	Campus LAN access (no internet required after initial setup)

Tip

MAC runs in CPU-only mode without a GPU. All AI features remain functional, though inference will be slower. For production use with 50+ concurrent users, an NVIDIA RTX 3060 12 GB or better is strongly recommended.

1.2. Installation

1.2.1. Step 1: Clone the Repository

git clone https://github.com/mbmuniversity2026/MAC.git
cd MAC

1.2.2. Step 2: Configure Environment

cp .env.example .env

Edit the .env file and set at minimum:

MAC_SECRET_KEY=<your-random-64-char-hex-string>
MAC_ENV=development

Important

The MAC_SECRET_KEY is used for JWT token signing. Generate a strong random key using: python -c "import secrets; print(secrets.token_hex(32))"

1.2.3. Step 3: Start the Platform

Windows (recommended):

start-mac.bat

Linux / macOS:

docker compose up -d

This starts all services:

mac-api – FastAPI backend
mac-nginx – Reverse proxy + static frontend
mac-postgres – PostgreSQL 16 database
mac-redis – Redis 7 cache
mac-qdrant – Vector database for RAG
mac-searxng – Self-hosted web search
mac-vllm-speed – GPU inference engine (if GPU available)
mac-whisper – Speech-to-text engine

1.2.4. Step 4: Open in Browser

Navigate to:

http://localhost

You will see the MAC login page:

1.2.5. Step 5: First Login

Use the development credentials to log in:

Role	Username	Password
Admin	`abhisek.cse@mbm.ac.in`	`Admin@1234`
Faculty	`raj.cse@mbm.ac.in`	`Faculty@1234`
Student	`21CS045`	`Student@1234`

1.3. Building the MBM Book Workspace Image

If you plan to use the MBM Book cloud IDE, build the workspace image:

docker build -t mbmbook-workspace:latest \
    -f docker/workspace/Dockerfile.lite docker/workspace/

This creates a ~6 GB image with runtimes for Python 3.11, Node.js 20, Java 17, Go 1.22, Rust 1.95, C/C++, Ruby, and PHP. Build time is approximately 5 minutes.

1.4. Stopping the Platform

Windows:

stop-mac.bat

Linux / macOS:

docker compose down

Note

Data is persisted in Docker volumes (pgdata, redisdata, qdrantdata). Stopping and restarting the platform does not lose any data.