Projects
CardTap
NFC AUTHENTICATION WEB APP
During my industrial placement I architected and developedCardTap, a Laravel‑based web application to authenticate worker qualifications via NFC‑enabled cards. I led all key development decisions, delivering a successful pilot with over 300 users in India and Bahrain, which led to the project’s approval for mass production.
The UI is simple and straightforward, with minimal complex Javascript. This was a deliberate design consideration as this website would often need to load and run by resource-constrained devices. In addition the target user of the app was not expected to be technologically literate so it had to have an intuitive design with minimal complex interactions.
Frontend: Custom built HTML, CSS, Javascript
Backend: Laravel Site using PHP
Hustl.io - Fantasy Stock Trading Website
BSc. First Year Team Project
For my First Year Team Project module we were tasked with designing and developing a web application of our choosing. In our team of seven we built hustl.io, a Django‑based website allowing users to trade with fake currency using real‑time stock data from the Finnhub API, add friend and create custom leagues. This introducing trading concepts without financial risk in an arcade‑style experience.
We managed real‑time stock price updates (approx. 100 per day per stock) on a live build, recording user transactions in a PostgreSQL database. I personally built the transaction, leagues and portfolio systems on backend; and the home screen, portfolio page and stocks page on the frontend.
Frontend: HTML, Bootstrap CSS, Javascript
Backend: Django using Python, with PostgreSQL database
Privacy‑Preserving Neural Network Inference via Homomorphic Encryption
BSc. Final Year Dissertation Project
My third year project involved developing a complex, privacy‑preserving machine learning system by reproducing a state‑of‑the‑art academic paper on bootstrapped CKKS encryption for a ResNet‑20 model.
This would involve translating theoretical algorithms from academic descriptions into a functional application, working independently without access to the original source code by synthesising information from multiple sources. I was able to build core components of a homomorphic encryption pipeline from scratch, including routines for encrypted convolution, noise‑reducing bootstrapping, and polynomial approximations for non‑linear ReLU activation functions.
I then conducted thorough analysis of the performance and accuracy of the system to evaluate how close this technology is to a mature solution, producing a high quality report using LaTeX. The report scored 73 and the project overall 71.
Technical Skills: C++ , LaTeX
PRINT & TAP (PAT)
AUTOMATED IMAGE GENERATION & DATABASE DESIGN
Designed and implemented a normalised database schema for the company’s card printing system to ensure data integrity and scalability.
Created a critical Python script to generate over 600 unique card images daily, delivering an MVP within one day to resolve a critical supplier issue.
Refactored the script into a fully automated cloud solution that sourced data from Onedrive, uploaded assets via an API, and sent automated error notifications.