NovaBuddy – AI-Powered Study Buddy

Serverless backend for the Amazon Nova AI Hackathon (Best Student App category).

Students upload lecture materials (PDFs, images) and NovaBuddy generates summaries, quizzes, flashcards, key concepts, and an interactive Q&A chatbot — all powered by Amazon Nova models via Bedrock.

Architecture

• Compute: 7 AWS Lambda functions (Python 3.12)
• API: HTTP API Gateway with Cognito JWT authorizer
• Auth: Amazon Cognito User Pool (email + password)
• Storage: S3 (files) + DynamoDB single-table (metadata + AI outputs)
• AI: Amazon Nova Lite via Bedrock Converse API
• IaC: Terraform

Prerequisites

1. Terraform >= 1.5 — install
2. AWS CLI configured with credentials that have admin access (or scoped to the services above)
3. Bedrock model access — enable amazon.nova-lite-v1:0 in the Bedrock console for us-east-1

Deploy

cd terraform

# Copy the example and adjust if needed
cp terraform.tfvars.example terraform.tfvars

terraform init
terraform plan
terraform apply

Terraform outputs the values you need:

Output	Description
api_url	Base URL of the HTTP API
cognito_user_pool_id	Cognito User Pool ID
cognito_client_id	Cognito App Client ID
s3_bucket_name	S3 uploads bucket name
dynamodb_table_name	DynamoDB table name

Terraform Variables

Variable	Default	Description
project_name	novabuddy	Prefix for all AWS resource names
region	us-east-1	AWS region (must support Nova models)
model_id	amazon.nova-lite-v1:0	Bedrock model ID for reasoning
embedding_model_id	amazon.nova-embed-multimodal-v1:0	Bedrock model ID for embeddings (stretch goal)
allowed_origins	["https://main.d28f2w8xvnnyj5.amplifyapp.com"]	Frontend URLs for API Gateway CORS and Cognito callback/logout URLs

See terraform/terraform.tfvars.example for a ready-to-use config file. For local dev, add e.g. "http://localhost:3000" to allowed_origins.

Create a Test User

# Replace <POOL_ID> and <CLIENT_ID> with Terraform outputs

# Sign up
aws cognito-idp sign-up \
  --client-id <CLIENT_ID> \
  --username [email protected] \
  --password YourPass123

# Confirm (admin shortcut — skips email verification)
aws cognito-idp admin-confirm-sign-up \
  --user-pool-id <POOL_ID> \
  --username [email protected]

# Get tokens
aws cognito-idp initiate-auth \
  --client-id <CLIENT_ID> \
  --auth-flow USER_PASSWORD_AUTH \
  --auth-parameters [email protected],PASSWORD=YourPass123

Save the IdToken from the response — use it as the Bearer token for all API calls.

API Reference

Base path: <api_url>/api/v1

All routes except /health require Authorization: Bearer <IdToken>.

Health Check

curl <API_URL>/api/v1/health

Upload a Lecture

# Step 1: Get presigned URL
curl -X POST <API_URL>/api/v1/upload/presigned \
  -H "Authorization: Bearer <TOKEN>" \
  -H "Content-Type: application/json" \
  -d '{"filename": "lecture1.pdf", "contentType": "application/pdf"}'

# Step 2: Upload file directly to S3 using the returned URL
curl -X PUT "<PRESIGNED_URL>" \
  -H "Content-Type: application/pdf" \
  --data-binary @lecture1.pdf

Process a Lecture (generate study aids)

curl -X POST <API_URL>/api/v1/lectures/<LECTURE_ID>/process \
  -H "Authorization: Bearer <TOKEN>" \
  -H "Content-Type: application/json" \
  -d '{"title": "Machine Learning Week 3"}'

This triggers Nova to extract text, generate a summary (300-500 words), 8-12 quiz questions, and key concepts. Takes 30-90 seconds depending on file size.

List Lectures

curl <API_URL>/api/v1/lectures \
  -H "Authorization: Bearer <TOKEN>"

Get Lecture Details

curl <API_URL>/api/v1/lectures/<LECTURE_ID> \
  -H "Authorization: Bearer <TOKEN>"

Chat about a Lecture

curl -X POST <API_URL>/api/v1/lectures/<LECTURE_ID>/chat \
  -H "Authorization: Bearer <TOKEN>" \
  -H "Content-Type: application/json" \
  -d '{
    "message": "Explain backpropagation in simple terms",
    "history": []
  }'

Delete a Lecture

curl -X DELETE <API_URL>/api/v1/lectures/<LECTURE_ID> \
  -H "Authorization: Bearer <TOKEN>"

Tear Down

cd terraform
terraform destroy

Project Structure

├── terraform/
│   ├── main.tf                    # Provider, random suffix
│   ├── variables.tf               # Input variables (incl. allowed_origins)
│   ├── terraform.tfvars.example   # Example variable values
│   ├── outputs.tf                 # API URL, Cognito IDs, bucket name
│   ├── cognito.tf                 # User Pool + App Client (callback/logout URLs)
│   ├── storage.tf                 # S3 + DynamoDB
│   ├── api_gateway.tf             # HTTP API, JWT authorizer, CORS, routes
│   └── lambdas.tf                 # IAM, Lambda functions, Layer, permissions
├── lambdas/
│   ├── layer/python/shared/       # Shared utilities (Lambda Layer)
│   │   ├── response.py            # HTTP response helpers + @api_handler
│   │   └── bedrock.py             # Bedrock Converse API wrapper
│   ├── health/                    # GET  /health
│   ├── get_presigned/             # POST /upload/presigned
│   ├── process_lecture/           # POST /lectures/{id}/process
│   ├── list_lectures/             # GET  /lectures
│   ├── get_lecture/               # GET  /lectures/{id}
│   ├── chat_lecture/              # POST /lectures/{id}/chat
│   └── delete_lecture/            # DELETE /lectures/{id}
├── docs/
│   └── ARCHITECTURE.md            # AWS architecture diagram + data flows
├── .gitignore
├── PRD.md
└── README.md