studypilot-aws

StudyPilot — AI Study Companion

Alexa-style AI study companion built with Amazon EC2, Apache, Amazon Bedrock, Amazon S3, IAM, and EBS.

Students paste their own lecture notes and receive:

• A 5-point revision summary + 90-minute study plan (Revise today)
• Five practice questions with answers (Make quiz)
• A plain-English explanation of the hardest concept (Explain simply)

Notes are stored privately in S3. The app runs on EC2 with Apache reverse-proxying the Node.js backend.

---

Architecture

Student → Classic Load Balancer (HTTP :80)
              ├── EC2 Instance 1 (Apache + Node.js)
              └── EC2 Instance 2 (Apache + Node.js)
                        │
                        ▼
                   Amazon Bedrock
                        │
              ┌─────────┴─────────┐
              ▼                   ▼
         Amazon S3            Amazon EBS
     (private notes)     (persistent disk)

Security

• Port 80 open to the world (Apache)
• Port 22 (SSH) restricted to your IP
• EC2 IAM role grants bedrock:InvokeModel and s3:PutObject only — no AWS keys in code

---

Project Structure

studypilot-aws/
├── frontend/
│   ├── index.html
│   ├── style.css
│   └── app.js
├── backend/
│   ├── server.js
│   ├── bedrock.js
│   ├── s3.js
│   └── package.json
├── scripts/
│   └── user-data.sh
├── docs/
├── LICENSE
└── README.md

---

Quick Start (Local Development)

1. Backend

   cd backend
   npm install
   cp .env.example .env   # create and fill values
   npm start
   

2. Frontend
   Serve the frontend/ folder with any static server, or open index.html after pointing API calls to http://localhost:5000.

---

AWS Deployment (EWB-style)

1. Create GitHub repo

Push this repository to https://github.com/YOUR_USERNAME/studypilot-aws.

2. Launch EC2

| Setting | Value | |—————-|————————————————| | Name | studypilot-server-1 | | AMI | Amazon Linux 2023 | | Instance type | t2.micro / t3.micro | | Key pair | studypilot-key.pem | | Root volume | 8 GB (EBS) | | Security group | SSH 22 from My IP · HTTP 80 from Anywhere |

Paste the contents of scripts/user-data.sh into User data (replace YOUR_USERNAME and bucket name).

3. IAM Role

Attach an EC2 instance role with:

{
  "Version": "2012-10-17",
  "Statement": [
    {
      "Effect": "Allow",
      "Action": ["bedrock:InvokeModel"],
      "Resource": "*"
    },
    {
      "Effect": "Allow",
      "Action": ["s3:PutObject"],
      "Resource": "arn:aws:s3:::YOUR_PRIVATE_BUCKET/notes/*"
    }
  ]
}

4. Private S3 Bucket

aws s3 mb s3://studypilot-notes-yourname --region ap-south-1

Keep Block Public Access enabled.

5. Bedrock Model Access

In the Bedrock console, request access to Claude 3 Haiku (or the model ID you set in .env).

6. Apache Reverse Proxy

The user-data script creates /etc/httpd/conf.d/studypilot.conf.
If you configure manually:

<VirtualHost *:80>
    DocumentRoot /var/www/html
    ProxyPreserveHost On
    ProxyPass /api/ http://127.0.0.1:5000/api/
    ProxyPassReverse /api/ http://127.0.0.1:5000/api/
</VirtualHost>

7. Test

Open http://YOUR_EC2_PUBLIC_IP (or Elastic IP / Load Balancer DNS).

---

Load Balancer (optional, after core demo works)

1. Launch a second identical EC2 instance.
2. Create a Classic Load Balancer:
   • Listener: HTTP 80 → instance port 80
   • Health check: /index.html or /health on port 80
   • Register both instances

---

Environment Variables

Variable	Description	Example
PORT	Backend port	5000
AWS_REGION	AWS region	ap-south-1
BEDROCK_MODEL_ID	Bedrock model ID	anthropic.claude-3-haiku-20240307-v1:0
S3_BUCKET	Private notes bucket	studypilot-notes-yourname

---

Cost-Safe Cleanup

• Stop / terminate EC2 when finished
• Release unassociated Elastic IPs (they incur hourly charges)
• Delete test objects from S3
• Delete unused EBS volumes

---

Submission Notes

EWB Project Title
StudyPilot — AI Study Companion Deployed on AWS EC2 with S3 and Bedrock

Devpost Tracks

• Primary: Alexa+ (simulated experience)
• Mini challenge: AWS Builder

---

License

MIT

This site is open source. Improve this page.