Built an AI chatbot agent with an intelligent model selector that routes prompts to the most suitable LLM

Integrated Chain-of-Thought prompting and a LangChain RAG pipeline for higher-fidelity answers

Delivered real-time dashboards with multiple interactive data-visualisation widgets

Designed a microservice architecture and automated deployment via GitHub Actions

Technology Stack:

• Backend: Node.js, Express.js (RESTful APIs, Jest unit tests)  

• Frontend: React.js (Responsive UI), TypeScript  

• AI / ML: OpenAI / Anthropic LLMs, dynamic routing, LangChain RAG  

• Security: JWT-based authentication & authorization

• Cloud & DevOps: AWS CDK (EC2, S3, RDS), Docker, GitHub Actions CI/CD 

• System Architecture: Microservices with event-driven messaging for real-time scalability

[Website Link]

Generates recipe ideas from pantry items and dietary preferences via OpenAI GPT

React/Vite SPA for ingredient management, saved/cooked tracking, and community sharing

Express backend handles session cookies, username-only auth, and admin moderation tools

Technology Stack:

• Backend: Express.js, Node.js

• Frontend: React.js SPA (Hooks, Context API)

• Real-Time: HTTP polling, i18n (react-i18next) 

• Security: JWT auth, HTTPS, input sanitisation

• Database: MongoDB (indexes tuned for chat workload)

• DevOps: Docker, GitHub Actions CI/CD

[GitHub Repo]

Implemented real-time messaging with a polling-based API and stateless JWT sessions  

Designed an i18n-ready React interface with context-aware filtering  

Structured user / message logic using clean OOP models on the server  

Technology Stack:

• Backend: Express.js, Node.js

• Frontend: React.js SPA (Hooks, Context API)  

• Security: JWT auth, HTTPS, input sanitisation

• Database: MongoDB (indexes tuned for chat workload)

• DevOps: Docker, GitHub Actions CI/CD

• Hosting: Render / AWS

[GitHub Repo]

Added Spring-Security RBAC so each healthcare role (doctor, insurer, public health, etc.) sees only its own workflows  

Built a dynamic Java Swing UI that renders components conditionally by role  

Created a medical-data tracker in Spring Boot + MySQL, tuning indexes and schemas for faster queries  

Implemented transactional messaging between roles for vaccine approvals and claims processing  

Technology Stack:

• Backend: Spring Security, Spring Boot

• Frontend: Java Swing

• Database: MySQL (indexed, normalised)

• DevOps: Maven, Git, CI/CD

[Project Visualisation Slides]

Scraped Petfinder listings into a Django database and exposed them via REST APIs  

Built interactive maps / charts that surface local adoption trends for Seattle users  

Trained lightweight ML models (linear & tree-based) to suggest high-priority matches  

Delivered a single-page HTML/JS interface for real-time search and filtering

Technology Stack:

• Backend: Django REST Framework, Python

• Data Scraping: requests, BeautifulSoup

• ML / Analytics: pandas, numpy, scikit-learn (Linear Regression, Decision Tree, Random Forest)

• Frontend: HTML, JavaScript (fetch API)  

• DevOps: Docker, GitHub Actions CI/CD  

[Project Summary]

[Project Visualisation Slides]