SmartDocs Architect

Final Project for EMGT 308 - Solutions Architecture and the Cloud

Project Overview

SmartDocs Architect is a cloud-hosted AI-powered document intelligence system built using Retrieval-Augmented Generation (RAG).

It allows users to upload documents, convert them into vector embeddings, retrieve semantically relevant context, and ask grounded questions based only on uploaded content.

The current MVP is deployed on AWS EC2 as a single-instance cloud workload.

Live Demo

Cloud Deployment (AWS EC2): http://54.226.223.159:8501/

Course Alignment

This project demonstrates core EMGT 308 concepts:

• Cloud deployment on AWS EC2
• Workload hosting in cloud infrastructure
• Layered solution architecture
• Infrastructure and application separation
• Scalable architecture planning

Features

• Upload PDF and TXT files
• Parse and chunk document text
• Generate embeddings using Gemini API
• Store vectors in ChromaDB
• Ask grounded questions
• Display retrieved source chunks
• Clear knowledge base

Implemented Architecture

User
↓
Internet
↓
AWS EC2 Instance
↓
Streamlit Application
↓
Gemini API + ChromaDB

Proposed Scalable Architecture

User
↓
Application Load Balancer
↓
Auto Scaling Group
↓
Multiple EC2 Instances
↓
Amazon S3

Technology Stack

• Python
• Streamlit
• Gemini API
• ChromaDB
• PyPDF
• python-dotenv
• AWS EC2

Project Structure

SmartDocs Architect/
│
├── app.py
├── requirements.txt
├── README.md
├── .gitignore
├── .env.example
│
├── app/
│   ├── __init__.py
│   ├── ingest.py
│   ├── rag.py
│   ├── parsers.py
│   └── utils/
│       ├── __init__.py
│       └── helpers.py
│
├── docs/
│   └── architecture-notes.md
│
├── screenshots/
└── chroma_db/

System Flow

1. Upload document
2. Parse text
3. Chunk content
4. Generate embeddings
5. Store vectors
6. Ask question
7. Retrieve relevant chunks
8. Generate grounded answer

AWS Deployment

• Ubuntu EC2 instance
• Security Group:
  • SSH (22)
  • TCP 8501
• Streamlit served externally

Mac / Linux:

python3 -m streamlit run app.py --server.port 8501 --server.address 0.0.0.0

Windows:

python -m streamlit run app.py --server.port 8501 --server.address 0.0.0.0

py -m streamlit run app.py --server.port 8501 --server.address 0.0.0.0

Public Access

http://54.226.223.159:8501

Screenshots

Application UI

AWS EC2 Deployment

SSH Access

Streamlit Running

Video Demonstration

YouTube: https://youtu.be/MLLgwZVvE4Q

Limitations

• Single EC2 instance
• Local vector persistence
• No authentication

Future Improvements

• Elastic IP
• Load Balancer
• Auto Scaling
• Amazon S3 storage

Author

Wineel Wilson Dasari

License

MIT License