Local Vendor Automation (VendorConnect) is a real-time, location-aware digital platform that connects verified street vendors with nearby customers.
It provides vendors with a live digital presence while enabling customers to discover, track, and pre-book products from trusted vendors in real time.
This project also includes AI-powered automation features such as OCR-based authentication, image classification using transfer learning, and a custom RAG chatbot for smart query answering.
- 📍 Real-time vendor location tracking & nearby discovery
- ✅ Admin-verified onboarding with secure authentication
- 🔍 OCR-based document authentication using EasyOCR
- 🖼 Product image classification using Transfer Learning
- 💬 Custom RAG chatbot for vendor/product query answering
- ⚡ Built using Microservices Architecture (Spring Boot + Eureka + Feign)
- 🔔 Real-time notifications for customers and vendors
To digitally empower street vendors and local businesses by giving them a real-time, location-based digital identity and enabling direct, transparent, and instant connections with customers.
As an AI/ML Developer, my contributions focused on automation and intelligent features:
- Developed an OCR-based vendor authentication module using EasyOCR to extract and validate vendor document details.
- Built a product image classification model using Transfer Learning to automatically categorize vendor products.
- Implemented a custom RAG chatbot that answers customer queries using vendor/product data.
- Collaborated with backend and frontend team members to integrate AI features into the overall system.
📌 These features helped reduce manual verification effort, improved onboarding speed, and enhanced customer experience.
VendorConnect is a mobile-first, location-aware platform designed to modernize informal street commerce.
- Mark live availability and current location
- List and manage products (admin-verified)
- Update stock and pricing in real time
- Receive and manage customer pre-bookings
- Notify nearby customers when arriving at a new location
- Discover verified vendors nearby
- View live vendor locations on a map
- Browse real-time product availability
- Pre-book items before visiting
- Receive instant notifications when vendors arrive nearby
- Live location tracking
- Product listing & management
- Inventory and stock control
- Order and pre-booking management
- Availability status (Online / Offline)
- Multi-location operation support
- Real-time notifications
- Secure authentication
- Admin-verified onboarding
- Nearby vendor discovery
- Live vendor location tracking
- Product browsing with pricing
- Pre-booking functionality
- Real-time arrival notifications
- Vendor ratings & reviews
- Verified vendor access only
- Vendor verification & approval
- Category and product moderation
- User and vendor management
- Platform monitoring
- Analytics & insights
- Security and access control
To streamline vendor onboarding and authentication, an OCR module was implemented using EasyOCR.
What it does:
- Extracts text from uploaded vendor documents (ID proof / license / certificates)
- Automatically reads fields like name, ID number, etc.
- Helps reduce manual verification time for admins
Impact:
- Faster vendor verification
- Reduced manual effort in authentication process
- Improved onboarding automation
A product image classification model was built using Transfer Learning to categorize vendor products automatically.
Use case:
- Vendors upload product images
- System predicts the product category (e.g., fruits, vegetables, grocery items)
Benefits:
- Improves product listing speed
- Reduces manual categorization work
- Enhances customer browsing and filtering
A custom RAG-based chatbot was developed to answer customer questions based on platform data.
Example queries supported:
- “Which vendors sell tomatoes nearby?”
- “Show me vendors with best ratings”
- “What is the price of milk today?”
- “Which vendor is currently online near my location?”
How it helps:
- Smart and fast customer support
- More accurate responses than normal chatbot
- Improves engagement and user experience
- Vendors broadcast live location and available products to nearby customers.
- Customers pre-book items before visiting a vendor.
- Vendors send real-time alerts when moving to a new location.
- Customers receive instant notifications with pricing and availability details.
- Admin verifies vendors using OCR-extracted details.
- Product images are auto-classified to speed up listing.
- Continuous internet and GPS access required.
- Vendor onboarding may take time due to verification.
- Digital literacy challenges for some vendors.
- Battery and location accuracy management on mobile devices.
- React.js
- React Native
- JavaScript / TypeScript
- CSS
- Java (Spring Boot)
- Spring MVC
- Spring Security
- MySQL
- Redis
- Microservices Architecture
- Spring Cloud Eureka (Service Discovery)
- OpenFeign (Inter-service Communication)
- Python
- EasyOCR
- Transfer Learning (Pretrained CNN Models)
- Image Classification Pipeline
- Custom RAG Chatbot (Retrieval-Augmented Generation)
- Docker
- Git & GitHub
- Postman
- JUnit
- Mockito
- Add multilingual support for vendors and customers
- Improve OCR accuracy using preprocessing techniques
- Add recommendation system for customers
- Deploy AI models as scalable microservices
- Improve chatbot responses using reranking + evaluation pipeline
If you want to collaborate or have questions about this project, feel free to connect:
Ayush Singh
LinkedIn: https://www.linkedin.com/in/ayush-singh-aiml/