maint-ai_demo

MaintAI — Equipment Troubleshooting Assistant

Client Demo · Advanced Multimodal RAG · Industry Standard

AI-powered assistant that lets maintenance technicians ask technical questions and get accurate, cited answers from equipment manuals. Supports scanned PDFs, troubleshooting photos, and global web fallback.

---

🛠 Features

• Multimodal Analysis: Upload photos of machine errors or screens for instant troubleshooting context.
• Role-Based Access (RBAC): Secure Login for Admins (Uploads/Deletes) and Technicians (Chat-only).
• Multilingual OCR: Supports 12+ languages (ZH, JA, KO, EN, HI, etc.) via PaddleOCR.
• Global Search Fallback: If manuals don't have the answer, one-click search queries the web for industrial solutions.
• Hyper-Fast Inference: Running on Llama 3.3 70B via Groq's LPU engine for near-instant responses.

---

🏗 Technology Stack

---

🚀 Deployment Guide

Option 1: Quick Start (Local)

1. Environment: Create a .env file from .env.example and add your GROQ_API_KEY.
2. Dependencies:

   # Backend
   cd backend && pip install -r requirements.txt
   python -m uvicorn app.main:app --reload
   
   # Frontend (Separate terminal)
   cd frontend && npm install && npm run dev
   

Option 2: Docker / Hugging Face Spaces

This project includes a unified Dockerfile in the root and is pre-configured for Hugging Face Spaces.

docker build -t maintai .
docker run -p 7860:7860 --env-file .env maintai

---

💡 Demo Flow

1. Login: Choose a role.
   • Admin: admin / admin (Can upload/edit library)
   • Technician: technician / tech (Secure chat access)
2. Upload: Add an equipment manual in the Manuals tab.
3. Chat: Ask a question or paste a screenshot.
4. Web Search: If the manual is insufficient, click the Globe Icon to fetch live maintenance data.

---

📂 Project Structure

maintai-demo/
├── Dockerfile              Unified multi-stage build (HF Ready)
├── .gitignore              Protects project secrets & local DB
├── backend/
│   ├── requirements.txt    RAG, OCR, & Search dependencies
│   └── app/
│       ├── services/auth.py  JWT & Role-based logic
│       ├── services/rag.py   ChromaDB + Groq + Web Fallback
│       └── main.py           FastAPI Entrypoint (serves Static UI)
└── frontend/
    └── src/
        ├── App.jsx         Multimodal Dashboard UI
        └── lib/api.js      Auth-aware API client

---

🔒 Security & Privacy

• Local Processing: Manuals are stored and indexed locally (ChromaDB); data never leaves your infrastructure except for LLM inference.
• LOTO Warnings: The system is prompted to prioritize safety and Lockout-Tagout protocols.