⚙️ AssetMind

AI-Powered Industrial Asset Intelligence Platform

Predict failures before they happen. Query manuals in plain English. Manage every asset from one dashboard.

Demo · Features · Architecture · Quick Start · Results · Team

---

📌 The Problem

Unplanned industrial downtime costs billions annually — most teams still run reactive maintenance, fixing equipment only after it fails. AssetMind flips this to proactive, data-driven maintenance: predicting failures weeks ahead, and putting OEM manual knowledge one question away.

🎬 Demo

📺 Watch the 2-min demo

✨ Features

🔮 Failure Prediction	RandomForest + XGBoost models forecast failure probability & Remaining Useful Life
📚 RAG Copilot	Ask natural-language questions across 9 OEM manuals, answered with source citations
🗂️ Asset Registry	Unified view of 25 industrial assets with live health scores
📋 Work Orders & Incidents	Track, prioritize, and root-cause maintenance events
🕳️ Knowledge Gap Detection	Flags where documentation coverage is missing
📊 Live Dashboard	Real-time KPIs, health scores, failure timelines

🏗️ Architecture

React 19 (Dashboard · Asset Explorer · Copilot Chat)
              │  REST / HTTP
              ▼
     FastAPI  (/predict  /ask  /equipment  /dashboard  /insights)
       │                        │
       ▼                        ▼
 ┌──────────────┐      ┌────────────────────────────┐
 │  PostgreSQL  │      │   ML Engine + RAG Pipeline  │
 │  Assets ·    │      │  RandomForest (failure) +   │
 │  WorkOrders ·│      │  XGBoost (RUL) · ChromaDB   │
 │  Incidents   │      │  over 9 OEM manual PDFs     │
 └──────────────┘      └────────────────────────────┘

Flow: sensor/operational features → RandomForest (failure probability) + XGBoost (RUL in cycles) → asset health score & alerts. User queries → embedded → matched against ChromaDB manual chunks → grounded LLM answer with citations + knowledge-gap flag.

📊 Results

Model	Task	Key Metric
RandomForest	Failure classification	96.4% accuracy, 0.973 ROC-AUC, 79.4% recall
XGBoost	RUL regression	13.28 cycles MAE, 18.40 RMSE

Top drivers of failure risk: torque, rotational speed, tool wear. (Trained on a synthetic dataset covering 25 assets × 2 years of operational data.)

🛠️ Tech Stack

Frontend: React 19 · Vite · Tailwind · Recharts Backend: FastAPI · SQLAlchemy · PostgreSQL 16 AI/ML: scikit-learn · XGBoost · ChromaDB (RAG) · PyMuPDF

🚀 Quick Start

git clone https://github.com/tasneem38/AssetMind.git && cd AssetMind

# Backend
cd backend && python -m venv venv && source venv/bin/activate
pip install -r requirements.txt
cp .env.example .env   # set DATABASE_URL, CHROMA_PERSIST_DIR, SECRET_KEY
python scripts/seed_data.py
python scripts/ingest_manuals.py
python scripts/train_ai4i.py && python scripts/train_cmapss.py
uvicorn app.main:app --reload --port 8000   # → localhost:8000/docs

# Frontend (new terminal)
cd frontend && npm install
cp .env.example .env.local   # VITE_API_URL=http://localhost:8000
npm run dev   # → localhost:5173

⚠️ Local proof-of-concept only — no auth on backend endpoints. Don't expose publicly without adding authentication.

📡 Core API

Endpoint	Purpose
POST /predict/failure	Failure probability (RandomForest)
POST /predict/rul	Remaining Useful Life (XGBoost)
POST /ask/copilot	Natural-language Q&A (RAG + relational data)
GET /dashboard/	Aggregate KPIs
GET /insights/knowledge-gaps	Documentation gap detection

Full interactive docs at /docs (Swagger UI) once running.

🗺️ What's Next

Real-time WebSocket alerts · IoT/MQTT sensor ingestion · model drift detection & auto-retrain · PDF/Excel report export.

👥 Team — CREW 1.0

Name	Role	GitHub
Tasneem Banu	Full-Stack Dev & AI/ML	@tasneem38
KS Bande Nawaz Ahamed	Backend & Data Architecture	@nawazks72

---

Built for smart industrial operations · Portfolio · LinkedIn · MIT License