This project refers to the Portfolio website you are currently viewing. I developed this website using React JS and Tailwind CSS. The website is hosted on Vercel, with updates being managed and pushed through GitHub.

The website trainwithstelios is a straightforward tool I designed for a personal trainer. It was crafted using React JS for the interactive features and Tailwind CSS for its clean look.

Once a user books a session, the details are automatically saved to a Google Sheets document and both the client and the personal trainer receive an email notification. The website is kept up-to-date by making changes through GitHub and is hosted on Vercel. In essence, it's a simple and efficient platform where clients can easily connect with the personal trainer.

This project was completed as part of the final thesis of my graduation. This was a cooparation of a team of 5 people and I was responsible for creating and gathering the data of the virtual patients and training 2 neural networks. I was also responsible for tranfering the models to the android by creating tflite versions of the models. I was also the lead programmer so I was responsible for coordinating the team in all programming related subjects

The first neural network was responsible for predicting the blood glucose levels for the next hour of the virtual patients and the second neural network was responsible for calculating the appropriate insulin bolus and basal dosage.

I used the Keras framework in Python to train the neural networks. The glucose prediction neural network was a LSTM network and I used the Adam optimizer and the Mean Squared Error loss function. The insulin dosage calculation neural network was also an LSTM network that used some of the weights from the first model to make the training faster. The insulin dosage calculation neural network was trained using reinforcement learning to calculate the insulin dosage.

The glucose prediction model achieved an RMSE of 10mg/dl±3 which based on the research conducted is very good result,and considering that the model was trained on a small dataset of 23 virtual patients and that I had to keep the model as small as small as possible so it could run on Android devices The insulin calculation model was able to keep virtual patients in the optimal range for more than 70% which is the recomended TIR based on the consensus in the field.

This project was the final project for the course ECE464. We had to create an application of our choise. I decided to create HabitTrack.

HabitTrack is an application that helps you track your habits. You can create habits and track them daily. It can send notifications to remind you to track them and has a counter that shows you how many times you tracked your habits. It can also help you one time tasks and it can send notifications to remind you to do them. The application uses firebase to store the data and it uses the firebase authentication to authenticate the users. HabitTrack was created using Java and Android Studio.

This project was the final project for a course. The course required that my team and I go to a local business and create a software for them . Our team decided to go to a local store that had a need of Digital Point of sale system. Our cooparation lead to the creation of this project and the customer was very pleased with the result. Our team was split into 5 roles and I was the programmer and I cooparated with the other team member to get the specification and requirements the customer needed.

The system can be used to create new products, save customers, add suppliers, add suppliers, create orders, create invoices, create receipts and create reports. This system can also create vouchers for customers to use. The vouchers are split in to 2 different types. The first type is a voucher that gives a discount to the customer and the second type is refund voucher. If a customer decided to return a product the system can create a refund voucher for the customer to use. Then the customer can exchange it for any other product in the store matching the value of the voucher. Alternatively, the customer can pay the difference, these can all be handled by the system. The system can export all the data in a pdf format like it that was requested by the customer or print it directly on paper.

I used Qt Creator and C++ to create the application. The database was created using MySQL and phpMyAdmin.