Building a Cloud-Native, AI-Powered Wealth & Budget Companion

In today's fast-paced digital economy, managing personal finance remains a major challenge for many. Traditional budgeting tools are either too generic, visually clunky, or lack actionable intelligence.

To bridge this gap, I built Expense Tracker—a premium, full-stack personal finance application. It combines high-end user experiences, Generative AI diagnostics, and a production-ready, cloud-native DevOps infrastructure.

✨ Key Features & User Experience

1. High-Fidelity UI & Micro-Animations

Smooth Transitions: Route changes trigger double-layer curtain sweeps via **GSAP** to mask load latencies.

Momentum Scroll: Lenis provides physical-mode scrolling, improving the overall page navigation feel.

Data Visualization: Interactive category distributions and balance overviews are rendered using responsive **Recharts** charts.

2. Generative AI Financial Companion

The platform integrates the Google Gemini API (`gemini-3.6-flash`) directly into the dashboard.

Smart Analysis: The backend compiles monthly income, expenses, and savings rates.

Custom Insights: Gemini generates personalized budget tips based on category spending variance.

Deterministic Fallback: A local rules engine executes heuristic checks to serve tips if API limits are reached.

3. Reporting & Enterprise Utilities

Excel Reports: Users can download formatted `.xlsx` spreadsheets generated on-the-fly using `ExcelJS`.

Profile Management: Dynamic image upload utilizing Multer middleware lets users update avatars instantly.

🚀 Cloud-Native DevOps & Infrastructure

The application was built with a DevOps-first mindset to handle real-world traffic patterns, ensuring scalability, automated delivery, and high availability.

📦 Containerization with Docker

The project is containerized using multi-stage Docker builds. The frontend uses an Nginx base image to serve optimized static assets, minimizing image sizes, while the backend uses a node-alpine layer.

☸️ Kubernetes (K3s) Orchestration

The app is deployed to a lightweight K3s cluster hosted on an AWS EC2 instance:

Horizontal Pod Autoscaling (HPA): Pods automatically scale (from 2 to 5 replicas) based on target utilization metrics of 70% CPU and 80% Memory.

Traefik Ingress Controller: Routes `/api` traffic to the backend and `/` to the Nginx-served frontend on a unified port.

Resource Limits: Configured to guarantee system stability and prevent resource contention.

🤖 CI/CD Automation with GitHub Actions

Whenever code is pushed to the `main` branch, an automated workflow executes:

1. Installs dependencies and runs syntax/linting checks.

2. Compiles production assets and builds Docker container images.

3. Publishes images to the GitHub Container Registry (GHCR).

4. Connects to the AWS EC2 instance over SSH, applies the Kubernetes manifests, and checks deployment readiness via `kubectl rollout status`.