Technical Architecture

---

Backend

---

• Framework: FastAPI for RESTful API endpoints (/chat, /reply, /end_chat).
• Server: Uvicorn for high-performance ASGI server implementation.
• AI Framework: CrewAI with four agents (Sentiment Analysis, Sentiment Review, Response Generation, Polishing) powered by Gemini 2.0 Flash and Flash-Lite.
• Tools: SerperDevTool for web searches to include contemporary fictional character references in positive/neutral responses (e.g., “Like Miles Morales, you’ve got great taste!”).
• Deployment: Docker containerized application hosted on Render for scalability and reliability.

Frontend

---

• Technologies: HTML, CSS, JavaScript (transitioned from React for simplicity).
• UI Features: Clean, cheerful interface with a chat widget, tabbed navigation (Home, Messages), and responsive design for mobile and desktop.
• Enhanced UX: Delayed message bubbles and WhatsApp-like input behavior for a natural, engaging experience.

Data

---

• Dataset: Amazon US Customer Reviews Dataset (Kaggle, 449,172 records, 148 product categories).
• Preprocessing: Sampled 2,000 reviews per rating (1–5 stars), cleaned text (removed HTML tags, stopwords, non-alphabetic characters), normalized (lowercase, standardized formats), and performed tokenization/lemmatization.
• EDA: Identified sentiment patterns (1–2 stars: negative, 3: neutral, 4–5: positive) and lexical trends via word clouds (e.g., “great” in positive, “delay” in negative reviews).

Setup Instructions

---

Prerequisites

1. Python 3.8+
2. Docker
3. Git
4. Render account (for deployment)
5. API Keys:
   • Serper API (SERPER_API_KEY)
   • Gemini API (GEMINI_API_KEY)

Local Setup

Clone the Repository

git clone https://github.com/<your-username>/amazon-chatbot.git
cd amazon-chatbot

## INSTALL DEPENDENCIES
```bash
pip install -r requirements.txt

## Set Environment Variables
### Create a .env file or export variables:

```bash

export SERPER_API_KEY=xyz
export GEMINI_API_KEY=xyz
# Place gen-lang-client-0184211067-8d635d347db2.json in the project root.
Run the Application

**uvicorn app:app --host 0.0.0.0 --port 8000**
**Access Locally**
**Open http://localhost:8000 in a browser.**

# Deployment on Render
## Create Render Account
## Sign up at render.com.

### Create New Web Service
### Link your GitHub repository (amazon-chatbot).

## Configure:

## Runtime: Docker

## Dockerfile: Use the provided Dockerfile in the repository root.

## Environment Variables:

```bash
**SERPER_API_KEY=7142a72718**
**GEMINI_API_KEY=AIzaSyCpHmrgHWrbiv3mow**
**Secret File: Upload gen-lang-client-0184211067-8d635d347db2.json as a secret file.**

## Deploy
Trigger a manual deploy from the Render dashboard.

Monitor logs for “Application startup complete” and ensure there are no errors.

## Access Deployed App
Visit: https://amazon-chatbot.onrender.com

**# Project Structure**


```bash
amazon-chatbot/
├── app.py                  # FastAPI application
├── agent_checkpoint.py     # CrewAI multi-agent logic
├── requirements.txt        # Python dependencies
├── static/
│   └── index.html          # Frontend HTML/CSS/JS
├── Dockerfile              # Docker configuration
└── gen-lang-client-*.json  # Google Cloud credentials
\