Full Stack Developer - from neural networks, to virtual reality RPGs.
In my free time, I like to make games and web apps, and work on other cool projects. I've compiled a list of them here that I'll be updating as I make more things in the future.
I am experienced in:
I've also worked with:
I can use:
I built a neural network in Python that learned to aim on the popular rhythm game osu! using Python. The entire project was livestreamed to Twitch almost 24/7 and quickly became a success after gaining over 150,000 views in 30 days. The neural network itself, dubbed "Neuro-sama", developed over time, starting off as a simple self-made feed forward neural network and eventually becoming a large convolutional neural network powered by Tensorflow. This was my first large project with a neural network but will certainly not be the last. You can watch a video showcasing the neural network here.
Abandoned Archive started as an entry into the Ludum Dare game jam competition, and now has a free demo available on Steam ahead of the 2022 early access launch. You can find more details on the Abandoned Archive website.
The player's journey begins in a medieval village terrorized by a multitude of different terrifying enemies that are led by a fearsome dragon. The player can converse with NPCs around the village by simply shaking their hand and receive quests through the games questing system. They can cast spells using gesture recognition with the wand, fight multiple monsters in real-time combat using the trajectory predicting bow and the infamous Sword of Light that the prophecy foretold of. The player must increase the power of the Sword of Light using the souls of enemies until they can raise an army against the dragon and eventually potentially save the village.
Click on the games to find out more.
DeltaFire is an ambitious multiplayer first person shooter and the successor to the award winning Scoped. It is being developed in Unity and uses Photon Networking for the multiplayer API and hosting. The game currently features 3 maps, 1 game mode, fully animated character models, 3 animated weapons and incredible graphics.
Casino Conquest is an MMO city-building game about raking in the most cash. Set in an infinitely expansive unorthodox world where casinos make the world go round, you must outbuild all other players vying to become #1.
Escape Earth was developed for Ludum Dare 42, a game jam in which you have 72 hours to make a complete game from scratch. The theme for Ludum Dare 42 was "Running out of space". You can play the game here: https://vedal.itch.io/escape-earth
Compulsive Casino is a captivating story-driven game created in just 7 days for the Community Game Jam 2019. You play as Thomas, who heavily enjoys the thrill of gambling, and follow his story as he plays through many questionable casino games. This game is available to play at https://vedal.itch.io/compulsive-casino
CircleFire was developed for Ludum Dare 41, a game jam in which you have 72 hours to make a complete game from scratch. The theme for Ludum Dare 41 was "Two incompatible genres". You can play the game here: https://vedal.itch.io/circlefire
Click on the projects to find out more.
I was a finalist for the Climate Hack.AI competition 2022, where I achieved the top score in the UK.
For this competition, we predicted future photovoltaic power in the UK using a neural network to help fight climate change.
Using Python I created a decision tree based Artificial Intelligence that learns by itself, with about 10 minutes of training it becomes unbeatable meaning it is only possible to lose or draw to this bot as it makes perfect decisions. These decisions are based on thousands of games that it has played with itself in order to slowly improve.
Using Unity and C# I created a bot that learns to play a 3D version of Pac-Man that I created. It was inspired by this video and after several hours of training it became better than me at the game.
Using Python I created a decision tree based Artificial Intelligence that learns by itself to play Connect 4. I left this training for several days until the model size was several gigabytes big, this was due to the vast number of possible outcomes in Connect 4. By this time the bot had become moderately good at Connect 4 but was not perfect as it had not been training for long enough. The entire event was streamed at Twitch for viewers to watch.
Alice is a voice controlled AI that I created using Python, a Rasperry Pi and some programmable LED lights. The AI is always listening and wakes when hearing "Alice" to respond to many commands that mostly control the lighting in my room. This means I can say "Alice lights rainbow" for a cool RGB rainbow theme in my room. The bot also interacts with my computer through a web server so that I can send commands from a python terminal on my computer or Alice can control specific elements on my computer like applications open or turn itself off.
Inspired by the likes of TwitchPlaysPokemon I created a python program that listens to my twitch chat and interprets commands like "A" or "Right 5" in order to play Dragon Quest IX through the desmume emulator. The twitch chat ended up surprisingly getting quite far into the game and it was cool to see a small community work together like that.