HouseScanner

AI-powered house inspection application with computer vision and intelligent checklist generation.

Screenshots

Tech Stack

• Frontend: React, TypeScript, Vite, TailwindCSS, PWA
• Backend: NestJS, Prisma, PostgreSQL
• AI/ML: FastAPI, LangChain, LangGraph, OpenAI GPT-4
• Storage: MinIO (S3-compatible), AWS S3
• Cache: Redis, ElastiCache
• Infrastructure: Kubernetes (Minikube, EKS), Helm, Docker, Terraform, Ansible
• Monitoring: Prometheus, Grafana

AI Workflow Architecture (LangGraph)

The AI agents service uses LangGraph to orchestrate a sophisticated 6-agent pipeline with parallel processing:

flowchart TD
    START([📸 Upload Images]) --> A1[🏠 Agent 1: House Classification]
    A1 --> A2[📋 Agent 2: House Checklist]
    A2 --> PARALLEL[🔄 Parallel Room Processing]
    
    PARALLEL --> A3[🚪 Agent 3: Room Classification]
    PARALLEL --> A4[✅ Agent 4: Room Checklist] 
    PARALLEL --> A5[🛋️ Agent 5: Products Analysis]
    
    A3 --> A6[⚖️ Agent 6: Pros/Cons Analysis]
    A4 --> A6
    A5 --> A6
    
    A6 --> END([📊 Final Report])
    
    classDef startEnd fill:#e1f5fe
    classDef agents fill:#f3e5f5
    classDef parallel fill:#e8f5e8
    
    class START,END startEnd
    class A1,A2,A3,A4,A5,A6 agents
    class PARALLEL parallel

Loading

Key Features:

• Parallel Processing: Rooms analyzed concurrently for 3-5x speedup
• LangSmith Tracing: Live workflow monitoring at smith.langchain.com
• Rate Limiting: Intelligent throttling (90K TPM, 500 RPM)
• State Management: Persistent checkpointing with memory
• Error Handling: Automatic retries with state preservation

---

Quick Start

Development (Docker Compose)

git clone https://github.com/InonELGABSI/houseScanner.git
cd houseScanner

# Setup environment files
cp client-pwa/.env.template client-pwa/.env
cp backend/.env.template backend/.env
cp agents-service/.env.template agents-service/.env

# Add your OpenAI API key to agents-service/.env
# OPENAI_API_KEY=your-key-here

# Start infrastructure and services
cd ansible && ansible-playbook dev-infra-setup.yml
cd .. && docker compose -f docker-compose.dev.yml up -d

Access: http://localhost:5173 (Login: housescanner@gmail.com / 12121212)

Development (Kubernetes)

# Start Minikube, build images, and deploy
make dev-setup

# Start port forwarding
make dev-ports

Access: http://localhost:3001

📖 Detailed instructions: DEV_SETUP.md

---

Production Deployment

Option 1: Docker on AWS EC2 (Ansible)

Simple deployment using Ansible to provision EC2 + RDS + S3.

cd ansible
ansible-playbook prod-all-steps.yml

Cost: ~$50-60/month

Option 2: Kubernetes on AWS EKS (Terraform)

Production-grade deployment with EKS, RDS, ElastiCache, and monitoring.

cd k8s/terraform/environments/prod
terraform init && terraform apply

cd ../../../
make prod-build && make prod-deploy

Cost: ~$150-200/month

📖 Detailed instructions: PROD_SETUP.md

---

Project Structure

├── client-pwa/          # React PWA frontend
├── backend/             # NestJS API
├── agents-service/      # FastAPI AI service (LangGraph)
├── ansible/             # Infrastructure automation
│   ├── dev-*           # Local development playbooks
│   └── prod-*          # AWS production playbooks
├── k8s/                 # Kubernetes manifests
│   ├── helm/           # Helm charts
│   ├── terraform/      # AWS infrastructure
│   └── environments/   # Dev/prod configs
├── scripts/             # Build and deployment scripts
└── .github/            # CI/CD workflows

---

Documentation

• DEV_SETUP.md - Complete development environment guide (Docker Compose + Kubernetes)
• PROD_SETUP.md - Production deployment guide (Ansible + Terraform/EKS)

---

License

MIT