🌩️ Cloud Computing (Fall 2025) – Assignment 1

🍽️ Dining Concierge Chatbot — AWS Serverless Application - YouTube Demo Link

A Dining Concierge Chatbot built using AWS serverless services, enabling users to receive personalized restaurant recommendations through a conversational interface.

---

👤 Student

Tilak Bhansali (tb3525)
MS in Computer Science @ NYU Tandon
🔗 GitHub: tilak30

---

🧭 Overview

This project implements a serverless, microservice-based web application built entirely on AWS.
The Dining Concierge Chatbot collects user dining preferences — location, cuisine, dining time, party size, and email — and provides restaurant recommendations via email.

It demonstrates how cloud-native services can be orchestrated to create an intelligent, scalable, and fully automated system. The project integrates multiple AWS components, including:

• Amazon S3 – to host the static frontend
• Amazon Lex – for natural language understanding and chatbot dialogue management
• AWS Lambda – for backend logic and event handling
• Amazon SQS – to decouple chatbot and recommendation processing
• Amazon DynamoDB – for restaurant and user session data storage
• Amazon SES – to send restaurant recommendations via email
• Amazon OpenSearch – for fast and efficient cuisine-based search
• Amazon CloudWatch – for monitoring and event-driven automation
• API Gateway – to expose REST endpoints connecting frontend and backend

Together, these services form an event-driven, serverless workflow that demonstrates real-world cloud computing principles.

---

☁️ Architecture

The chatbot is built entirely on AWS with the following services:

Service	Role
Amazon S3	Hosts the static frontend web application
Amazon Lex (V1/V2)	Conversational AI that handles user dialogue and slot collection
AWS Lambda (LF0, LF1, LF2)	Backend logic for API, Lex hook, and recommendation processing
API Gateway	REST interface connecting the frontend to the backend
Amazon SQS	Queue to decouple chatbot and restaurant recommendation services
Amazon DynamoDB	Stores restaurant data and user search history
Amazon OpenSearch	Enables fast restaurant search by cuisine or location
Amazon SES	Sends restaurant recommendations via email
Amazon CloudWatch / EventBridge	Monitors logs and triggers backend workflows

---

✨ Features

✅ Interactive chatbot interface using Amazon Lex
✅ Collects dining preferences through natural conversation
✅ Pushes user data to SQS for asynchronous processing
✅ Fetches 1000+ restaurants from the Yelp API and stores them in DynamoDB
✅ Indexes restaurants in OpenSearch for quick retrieval
✅ Sends personalized restaurant suggestions via SES
✅ Fully serverless, event-driven, and auto-scalable architecture

---

🧩 Components

🖥️ 1. Frontend (Web Client)

• Static web application hosted on S3
• Communicates with backend via API Gateway
• Sends user messages to the /chatbot endpoint connected to LF0

---

⚙️ 2. Backend Services

LF0 — Chatbot API

• Exposed via API Gateway
• Receives messages from the frontend and interacts with Lex

LF1 — Lex Code Hook

• Connected as both DialogCodeHook and FulfillmentCodeHook
• Implements intents:
  • GreetingIntent
  • ThankYouIntent
  • DiningSuggestionsIntent
• Validates slot data (email, party size, etc.)
• Sends collected user preferences to SQS

LF2 — Recommendation Service

• Triggered by SQS messages
• Queries DynamoDB and OpenSearch for restaurant matches
• Sends formatted email suggestions via SES

---

🔁 Data Pipeline

🍴 1. Yelp API Data Collection

• Retrieves restaurant data from Yelp Fusion API across 5 cuisines:
  Chinese, Indian, Italian, Mexican, and Thai
• ~200 restaurants per cuisine (~1000 total)
• Each record includes:
  • Business ID
  • Name
  • Address
  • Coordinates
  • Reviews
  • Rating
  • Zip Code
  • Cuisine
  • insertedAtTimestamp

💾 2. Storage

• Restaurant data stored in DynamoDB (yelp-restaurants table)
• Indexed in OpenSearch for quick querying

📬 3. Processing & Communication

• Lex → LF1 → SQS → LF2 → SES
• User requests flow through queue-based and event-driven architecture for scalability and reliability

---

💬 Example Conversation Flow

🗣️ User Input	🤖 Bot Response
Hi	Hi there! How can I help you today?
Find me an Italian restaurant	Sure, what location are you looking for?
New York	Great! For how many people?
2	Got it. What time would you like to dine?
7 PM	Please share your email so I can send recommendations.
abc@example.com	You're all set! I'll email you restaurant suggestions shortly.

---

🧠 Key Learnings

• 🧩 Designing serverless, event-driven cloud applications
• 💬 Integrating Lex, Lambda, SQS, DynamoDB, SES, and OpenSearch
• 🍴 Applying Yelp API for large-scale restaurant data ingestion
• ⚙️ Automating backend pipelines using CloudWatch and EventBridge
• 🔄 Understanding inter-service communication and scalability in AWS

---

🎥 Demo

📺 YouTube Demo: YouTube Demo Link

---

🏗️ Architecture Diagram

The diagram below illustrates the AWS serverless architecture of the Dining Concierge Chatbot.

---

⭐ Project by: Tilak Bhansali (tb3525)
📍 NYU Tandon School of Engineering – Cloud Computing (Fall 2025)

---

Data Flow (Mermaid)

graph LR
  User -->|1. Send message| API[API Gateway]
  API -->|2. Invoke| LF0[LF0 Lambda - Chatbot API]
  LF0 -->|3. Forward| Lex[Amazon Lex]
  Lex -->|4. Invoke| LF1[LF1 Lambda - Lex Hook]
  LF1 -->|5. Push| SQS[SQS Queue]
  SQS -->|6. Trigger| LF2[LF2 Lambda - Recommendation Service]
  LF2 -->|7. Query| DB[DynamoDB]
  LF2 -->|7b. Search| OS[OpenSearch]
  LF2 -->|8. Send Email| SES[Amazon SES]
  SES -->|9. Deliver| Inbox[User Inbox]

Loading