AI-powered inventory intelligence and seller analytics dashboard designed for fashion e-commerce businesses.
The platform helps online sellers monitor products, analyze sales performance, detect market trends, and receive AI-driven business recommendations across multiple marketplaces.
This project is being developed during a BTK hackathon as a smart analytics platform for fashion sellers.
The system combines:
- Multi-channel marketplace integration
- Product and sales analytics
- AI-generated business summaries
- Trend detection and monitoring
- Inventory intelligence
- Review analysis
- Smart seller recommendations
- Automated synchronization system
- Inventory and return monitoring
- Market intelligence alerts
Simulated integrations for:
- Trendyol
- Hepsiburada
- Amazon
The system architecture supports future real API integrations.
Users can connect marketplace accounts using API-key based mock integrations.
Each connected account supports:
- Platform-based account linking
- Source user identification
- Manual synchronization
- Last sync tracking
- Account activation/deactivation
- Multi-account management
Track product performance across multiple platforms:
- Product listing management
- Sales performance tracking
- Inventory monitoring
- Return monitoring
- Ratings and reviews analytics
- Platform-based product comparison
The platform converts complex seller data into simple business insights using LLM-powered analysis.
AI-generated reports include:
- Sales summaries
- Product performance analysis
- Inventory analysis
- Review insights
- Business recommendations
StockRadar includes a hash-based AI caching mechanism to avoid unnecessary LLM requests.
If the report data has not changed, the system reuses the previously generated AI response instead of sending a new request to the LLM.
This improves:
- Response speed
- Token efficiency
- System performance
- Cost optimization
Detects rising fashion trends and matches them with seller inventory.
Features include:
- Trending product detection
- Trend-based recommendations
- Market opportunity analysis
- Fashion trend monitoring
Analyzes customer reviews to identify:
- Negative feedback patterns
- Size-related issues
- Shipping complaints
- Product quality insights
- Customer sentiment
Aggregates fashion and e-commerce related news and summarizes them into seller-focused insights and alerts.
The system generates:
- Market alerts
- Trend alerts
- Opportunity notifications
- Industry intelligence summaries
Provides actionable AI suggestions such as:
- Increase stock
- Promote trending products
- Optimize pricing strategy
- Improve size charts
- Expand product categories
Supports:
- Manual synchronization
- Scheduled synchronization
- Multi-platform data refresh
- Import tracking
- Sync logging system
- Connected account management
Generates intelligent alerts for:
- Stock risks
- Trend opportunities
- Return issues
- Negative reviews
- Market changes
- Platform activity
Includes admin monitoring tools for:
- User management
- Connected account tracking
- AI cache management
- Platform activity monitoring
- System analytics
- React
- Vite
- Python
- FastAPI
- SQLAlchemy
- JWT Authentication
- APScheduler
- PostgreSQL
- Docker
- Docker Compose
- Marketplace simulation engine
- JSON-based marketplace mock sources
- Multi-platform mock integration system
- Gemini-powered report summarization(Gemini 2.5 Flash)
- AI-generated recommendations
- Trend analysis engine
- Review sentiment analysis
- Inventory intelligence system
Frontend (React + Vite)
↓
FastAPI Backend
↓
Router Layer
↓
Service Layer
↓
PostgreSQL Database
AI Layer:
- Gemini API
- AI Report Summaries
- AI Recommendations
- AI Review Analysis
- AI Stock Analysis
- AI Cache System
Marketplace Simulation:
- JSON mock sources
- API-key based account matching
- Product, order and review imports
## Installation
### Clone the project
```bash
git clone <repository_url>
cd project
cp .env.example .envFill the required API keys inside .env.
docker compose up --buildAfter the containers are running:
docker compose exec backend python scripts/generate_mock_data.py- JWT-based authentication
- Protected API routes
- Role-based admin authorization
- User-based marketplace isolation
- Connected account validation
- API-key based mock marketplace connection
Admin users are created using a seed script.
After the containers are running, open a new terminal and run:
docker compose exec backend python scripts/create_admin.pyThe script creates an admin user in the database.
Admin users can access protected admin routes such as:
/admin/users
/admin/summary
/admin/ai-cache
/admin/connected-accounts
/admin/listings
/admin/users/{user_id}
/admin/users/{user_id}/products/nested
Admin access is controlled by role-based authorization.
Normal sellers cannot access admin endpoints.
FastAPI Swagger documentation is available at:
http://localhost:8000/docs
- Create or login as a seller.
- Connect a marketplace account using an API key.
- Import product, order and review data.
- View inventory and sales analytics.
- Generate AI-powered reports.
- Review market news and trend insights.
- Use admin panel for system monitoring.
- Real marketplace API integrations
- Mobile application
- AI pricing optimization
- Seller performance scoring system
- Multi-language support
Developed by EastCoders.
Taha Buğra KÜÇÜKENEZ Ayla Shamsi Emrullah Gülseven