job-tracker-serverless

Current status: POC in progress

A production-style serverless web application built with React, AWS CDK, and AWS managed services to track job applications and application status.

Why I Built This

This project demonstrates a production-style serverless architecture using AWS CDK and managed services. It is designed to:

• Showcase infrastructure-as-code (IaC) practices
• Demonstrate authentication via JWT (Cognito)
• Model real-world CRUD patterns with DynamoDB
• Separate development and stable release branches

---

Architecture

flowchart LR
    User[User / Browser]

    CF[CloudFront]
    S3[S3 Bucket - React Static Site]

    Cognito[Cognito User Pool - JWT Tokens]

    APIG[API Gateway - JWT Authorizer]
    Lambda[Lambda - Node.js Backend]
    DDB[DynamoDB]

    User --> CF
    CF --> S3

    User --> Cognito
    Cognito --> User

    User --> APIG
    APIG --> Lambda
    Lambda --> DDB

Loading

---

Tech Stack

• Frontend: React + TypeScript
• Infrastructure: AWS CDK (TypeScript)
• Hosting: S3 + CloudFront
• Authentication: Cognito User Pools (JWT)
• API: API Gateway (HTTP API)
• Compute: AWS Lambda (Node.js)
• Database: DynamoDB

---

Local Development

Backend

cd backend
npm install
npm run dev

Runs on: http://localhost:3001

Frontend

cd frontend
npm install
npm run dev

Runs on: http://localhost:5173

---

Current Progress

• CDK project initialized
• API Gateway + Lambda deployed
• DynamoDB table created
• React app scaffolded
• Frontend connected to backend
• Cognito authentication integrated
• Deployed to CloudFront

---

Tech Debt / Suggestions

• appliedDate is only checked as a string; consider validating/parsing it (e.g., ISO date) to avoid inconsistent formats.
• Replace DEMO_USER_ID with an authenticated user id when integrating auth.
• Use a validation library (zod/joi) or request DTOs for clearer and more maintainable input validation.
• If the store becomes async (real DB), make handlers async and await persistence; also consider returning a Location header for the new resource.
• Consider an environment variable for the port (e.g., process.env.PORT ?? 3001), tighten CORS for production, add request-logging and error-handling middleware, and implement graceful shutdown for the server.
• Add reserved concurrency to Applications Lambda to cap cost
• Add caching strategy notes (CloudFront for frontend; API caching not needed yet) You can bridge that by adding one small AI feature:
• “Given a job description, extract keywords + generate tailored bullet points” (calls an LLM)
• Save results per application