LANGUAGES
TOOLS

Kshananth Jeyamohan

£20.06

About Me (˶ᐢ ᵕ ᐢ˶)

Hi! I'm a final-year Computer Science student at Newcastle University. I'm from London and plan to work there after I graduate.

I know I want to work in software engineering, but I haven't settled on a specialism within it yet, so I'm using my final year to try different areas.

Right now, I'm focusing on the fundamentals, practising LeetCode and building projects to bridge the gap between university theory and industry. I'm also applying for graduate software engineering roles.

The shelves on your right hold my projects, from uni coursework to things I'm still building. Click one to have a look! (◕‿◕)

改善 (Kaizen)
Turf
E-A
Kaizen Today

Kaizen Today is a personal project I'm currently working on. It's a chill, low-pressure social space where people can talk about their day-to-day lives, share how they're feeling, and work towards their longer-term goals through regular check-ins.

I'm building it because I've always felt that a lot of social media can be pretty toxic and overwhelming. I wanted to make something that feels more relaxed, somewhere you can just talk about the little things going on in your day without everything having to be polished, impressive, or performative.

Kaizen Today is being built with React, TypeScript, and Tailwind CSS, with an Express backend and PostgreSQL planned for the database.

Turf

Turf is a running game we built as a team of seven for a second year university module. Runners claim territory in their local area just by running through it in the real world, so a normal run turns into a game against other people nearby.

I built most of the backend, including starting and ending runs, live GPS tracking, and an anti-cheat system that checks how fast you're moving between GPS points. It rejects anything impossible, like someone driving instead of running, but allows for the odd bad GPS reading so honest runners don't get punished. I also wrote 14 automated tests to make sure it all worked.

Turf was built with Python, FastAPI and PostgreSQL on the backend, with a Flutter app on the frontend.

Image Reconstruction via Evolutionary Algorithms

For a second year module, I wrote an evolutionary algorithm that tries to recreate a painting using only 100 semi-transparent shapes. It starts with a population of random attempts, keeps the ones that look closest to the real painting, and mutates them over thousands of generations.

I improved it over seven versions, adding tournament selection, μ+λ elitism and mutations that get smaller as the picture gets closer, and tested each version over five seeded runs. Accuracy went from 85% in the first version to 96% by the end.

Image Reconstruction was built in Python, using Pillow to draw the shapes.