Selected projects
A selection of my most recent projects, from newest to oldest
Learning by building
Things I built explicitly for the purpose of deeper understanding.
Decision Tree from Scratch ✅ - [Python] A tutorial on implementing a decision tree from scratch.
Recurrent and fully connected neural networks from scratch ✅- [Python] Two notebooks where I develop two foundational machine learning models, the multi-layer perceptron and recurrent neural network, from scratch.
Card games No Thanks! and Take 5 with Q-learning ✅[Python] This project explores reinforcement learning strategies for card games No Thanks! and Take 5!, demonstrating the effectiveness of Q-learning with linear function approximation for complex game states.
Other Projects
Predicting psychiatric indications from EEG data ✅ - [Python] A package developed for the completion of Masters degree. Implements a full cleaning and machine learning pipeline for EEG data .
Print neural network parameters ✅ - [Julia] package to print parameters of a Lux neural network in a readable way.
Machine learning on EEG data for BCI applications in esports ✅ - [Python] Classification of player mental states during League of Legends play using electrical signals recorded from the scalp (EEG).
Auto-hyperparameter tuning for Reinforcement Learning with Optuna ✅- [Python] Learning the Gymnasium environment and hyper-parameter tuning for reinforcement learning on toy settings (Hopper and Bipedal walker).
Study of EEG recordings in linguistic experiments ✅[Python] This project examines how rhythm affects sentence ambiguity processing in the brain using EEG data from multilingual participants, with findings suggesting minimal impact for late German learners.
Observer - a python package for preparing and analyzing data from Dota 2 replays ✅- [Python] For the paper “Implicit Coordination Dynamics: A Synchrony-based Study on Team Positioning and Performance in Competitive Dota 2”, I developed a package to extract and analyze data from Dota 2 replays.