I like

• developing techniques to extract meaningful insights from text data.
• to explore word embedding models to understand how they capture meaningful semantics.
• to understand biases captured by language models.