I largely focus on three research areas, Reinforcement Learning (RL), Procedural Content Generation (PCG) and Natural Language Processing (NLP). In RL, my focus is on improving generalisation of agents to new scenarios, trying to reduce the brittleness of policies when there are slight differences between training and testing. In addition to this, I enjoy trying to apply RL to more tangible problems, notably our WITS-FC Robocup team. In NLP, I am interested in low-resourced languages, and how we can best leverage transfer learning from other languages to improve performance. Finally, in PCG, I generally like to focus on level generation, with an overarching aim of generating complex and high-quality structures with minimal human design.