A serverless, full-stack expense tracker built entirely on AWS — no servers to manage, no infrastructure to maintain. Add expenses, track spending by category, and view monthly reports, all backed by a pay-per-use backend.

🔗 Live Demo: https://roshni-expense-book.netlify.app

I wanted to understand serverless architecture from the ground up — not just read about Lambda and API Gateway, but actually wire them together, break things, read the CloudWatch logs, and fix them. This project is the result: a working expense tracker where every part of the stack — database, backend, API layer, and frontend — was built and debugged by hand.

┌─────────────┐ HTTPS ┌──────────────┐ Invoke ┌─────────────┐ Read/Write ┌──────────────┐ │ Frontend │ ─────────────▶ │ API Gateway │ ─────────────▶ │ Lambda │ ─────────────────▶ │ DynamoDB │ │ (HTML/CSS/JS)│ │ (REST API) │ │ (Python) │ │ (NoSQL) │ └─────────────┘ ◀───────────── └──────────────┘ ◀─────────────── └─────────────┘ ◀───────────────── └──────────────┘ Netlify Roshni stage 4 functions, Partition key: userId IAM execution role Sort key: expenseId

Request flow: the browser calls a REST endpoint → API Gateway (with Lambda proxy integration) forwards the raw event to the matching Lambda function → the function reads/writes DynamoDB using an IAM role scoped for that access → the response is serialized back through API Gateway to the browser, with CORS headers so the request isn't blocked cross-origin.

Layer Technology Frontend HTML5, CSS3, vanilla JavaScript (fetch API) API layer Amazon API Gateway (REST API, Lambda proxy integration) Compute AWS Lambda (Python 3.x) Database Amazon DynamoDB Access control IAM roles & policies Hosting Netlify (frontend) Version control Git & GitHub

Method Endpoint Description POST /expenses Add a new expense GET /expenses?userId={id} Get all expenses for a user DELETE /expenses/{expenseId}?userId={id} Delete a specific expense GET /report?userId={id} Get total spend + category-wise breakdown

Add expenses with amount, category, description, and date

View all expenses in a clean, ledger-style list

Delete individual entries

Auto-generated monthly report with category-wise totals

Fully responsive UI, no frameworks — just HTML/CSS/JS talking directly to the API

Building this surfaced real, non-obvious serverless issues that I worked through one by one:

Lambda proxy integration — without it enabled in API Gateway, event['queryStringParameters'] came back empty even though Postman showed the request going through fine.

IAM least-privilege — Lambda's default execution role only allows writing logs. Every DynamoDB call needed an explicit permission grant, which surfaced as AccessDeniedException in CloudWatch until fixed.

Decimal serialization — DynamoDB returns numeric values as Python Decimal, which json.dumps() can't serialize by default. Fixed with a custom default= handler.

CORS (two-part fix) — a browser blocks cross-origin calls entirely different from how Postman behaves. Fixing it required both a Access-Control-Allow-Origin header in every Lambda response and enabling CORS at the API Gateway resource level for the preflight OPTIONS request to succeed.

Case sensitivity — a path parameter named {Id} in API Gateway silently mismatched event['pathParameters']['id'] in Python, producing a KeyError that only made sense after reading the CloudWatch traceback.

expenses-book/ ├── index.html # Frontend (self-contained) ├── backend/ │ ├── addExpense/lambda_function.py │ ├── getExpense/lambda_function.py │ ├── deleteExpense/lambda_function.py │ └── getMonthlyReport/lambda_function.py └── README.md

The frontend is a single static file — no build step required.