🤖 DalFitBot

DalFitBot is a cloud-native chatbot application designed to provide intelligent, real-time answers to user queries about a facility using internal business data. Leveraging Retrieval-Augmented Generation (RAG) and powered by the LLaMA language model, DalFitBot delivers context-aware, accurate responses by retrieving relevant data chunks and combining them with generative AI capabilities.

🔧 Key Features

• RAG Architecture: Combines document retrieval with generation for precise, grounded answers.

• LLaMA-Based NLP: Utilizes Meta's LLaMA model for high-performance, contextual language understanding.

• FastAPI Backend: Built with Python and FastAPI for efficient, scalable, and asynchronous API operations.

• Cloud-Native Design: Fully deployed on AWS with high availability and security in mind.

☁️ AWS Cloud Infrastructure

• AWS Cognito – User authentication and secure access control.
• API Gateway – Handles RESTful API requests and routes traffic to backend services.
• Network Load Balancer (NLB) – Distributes traffic across EC2 instances with high throughput.
• EC2 Instances – Host the FastAPI application and inference service.
• NAT Gateway – Enables outbound internet access from private subnets.

📦 Tech Stack

• Language: Python
• Framework: FastAPI
• Model: LLaMA (integrated with a RAG pipeline)
• Deployment: AWS (Cognito, API Gateway, NLB, EC2, NAT)

Cloud Deployment

Prerequsites: AWS Account, Terraform

1. Set the aws credentials as the default credentials using aws configure. Refer AWS Docs if you need help.
2. Go to the root directory where the main.tf file is present and run the below command in sequence.

terraform init           # To initialize the project.
terraform validate       # To validate the terraform script and AWS Configuration.
terraform plan           # Displays the resources to be created.
terraform apply          # Start creating the resource in the cloud. 

The React Endpoint will be displayed through which the application can be accessed.