ZEN STORE

Zen Store — eCommerce Intelligence Platform

Real-time AI-powered customer support and automated agent escalation for global footwear retail.

---

🏆	Industrial Grade — Phase 1 Project Built as a high-performance demonstration of AI-integrated eCommerce and automated customer support orchestration.

---

Live Demo • Documentation • Report a Bug • Request a Feature

---

Table of Contents

1. Introduction
2. Key Features
3. Technology Stack
4. System Architecture
5. Directory Structure
6. Installation and Configuration
7. API Specification
8. License

---

Introduction

Zen Store is an industrial-grade eCommerce solution engineered for high-availability footwear retail. The platform bridges the gap between static product browsing and dynamic customer support by utilizing a custom-trained Sentence-Transformer model for real-time semantic inquiry resolution.

---

Key Features

Autonomous Customer Support (NLP)

• Semantic Vector Matching: Leverages all-MiniLM-L6-v2 for mapping user queries to a multidimensional knowledge space.
• Dynamic Product Awareness: Real-time synchronization with the product catalog (78+ SKUs) allows for accurate pricing and stock-level responses.
• Intent-Based Escalation: Automated frustration detection triggers a transition state from AI-handled to Human-Agent-waiting status.

Professional Frontend Engineering

• State-Driven UI: Real-time UI updates based on chat status (AI Active, Waiting, Live Agent).
• Sub-directory Routing: SEO-friendly routing structure with dedicated modules for Men, Women, and Sports collections.
• Unified Design System: A bespoke CSS framework utilizing CSS variables for consistent glassmorphism and typography across 15+ sub-views.

Scalable Backend Infrastructure

• Asynchronous Processing: FastAPI-driven service layer ensuring sub-200ms response times for AI inquiries.
• Data Persistence: Normalized SQLite database for logging interactions, session management, and agent replies.

---

AI Chatbot Implementation

AI-Powered Customer Support Service.

1. Functional Requirements

• FAQ Resolution: The system utilizes semantic similarity to resolve customer queries against a pre-defined FAQ dataset.
• Automated Escalation: Implements a "Human-in-the-Loop" protocol. When query confidence falls below a threshold or user frustration is detected, the system transitions to an escalation state.
• Interaction Logging: All user-bot interactions and escalation events are persistently logged in a SQLite database for audit and training purposes.

2. Non-Functional Requirements

• Response Latency: The system is optimized for sub-200ms inference times (local) and < 1.5s via remote API, exceeding the < 2s requirement.
• Accuracy Threshold: Semantic vector matching provides an estimated accuracy of > 85% for domain-specific inquiries.

3. Architecture & AI Layer

• Frontend: Responsive web interface developed with Vanilla JS and CSS.
• Backend: Python-based FastAPI service layer.
• AI Layer: Utilizes a pre-trained all-MiniLM-L6-v2 Sentence-Transformer model from Hugging Face for intent classification.
• Database: SQLite integration for persistent session and interaction storage.

---

Technology Stack

Component	Technology	Utility
Frontend	JavaScript (ES6+)	Application logic and state management
Styles	CSS3	Custom Design System (Glassmorphism)
API Framework	FastAPI	High-performance asynchronous backend
AI Model	Sentence-Transformers	Semantic intent and similarity matching
Database	SQLite3	Session and interaction persistence
Server	Uvicorn	ASGI server implementation

---

System Architecture

The platform follows a decoupled Frontend-Backend architecture, communicating via a RESTful API.

1. Client Layer: Dispatched via standard HTTP. Handles UI rendering and local state.
2. Logic Layer: FastAPI service processing NLP tasks and database I/O.
3. Data Layer: JSON-based product datasets and SQLite interaction logs.

---

Directory Structure

/ (Project Root)
├── index.html              # Primary entry point
├── 404.html                # Global error handler
├── assets/                 # Centralized resources
│   ├── css/                # Main stylesheet and design tokens
│   ├── js/                 # Modular application logic (Auth, Cart, AI)
│   └── images/             # Product and UI assets
├── pages/                  # Route-specific modules
│   ├── account/            # Authentication and Profile views
│   ├── admin/              # Management and Verification dashboard
│   └── [category]/         # Product collection views
└── core/                   # System-critical infrastructure
    ├── server/             # FastAPI source and ML models
    ├── scripts/            # Synchronization and maintenance tools
    └── docs/               # Technical specifications

---

Installation and Configuration

Environment Requirements

• Python 3.10+
• Node.js (Optional for advanced tooling)
• Git

Backend Deployment

1. Navigate to the infrastructure directory:

   cd core/server

2. Initialize virtual environment and dependencies:

   pip install fastapi uvicorn sentence-transformers torch

3. Initialize the service:

   python main.py

---

API Specification

Endpoint	Method	Description
/ask	POST	Submits a query to the AI engine for semantic matching
/status/{session_id}	GET	Retrieves current chat status (AI/Waiting/Agent)
/agent-reply	POST	(Admin) Injects a human agent response into the session

---

License

Project documentation and source code are licensed under the MIT License.

Author

Muhammed Anas GitHub: i-anasop