This thesis project aims to analyse the effect of human-Large Language Model (LLM) collaboration in reflective essay classification tasks. Reflective essays provide insights into personal growth and critical thinking, but evaluating them manually is subjective and time-consuming. By collaborating human coder and an advanced language model trained on extensive text data, this project aims to explore a human-LLM collaboration performance in classifying reflective essays. The model will be trained on a dataset of approximately 20 reflection scripts, starting with data preprocessing and followed by fine-tuning for the specific task of essay classification. Upon training, both human and the LLM will compare the result of each essays, along with its explanation, and iterate this step until both human and the LLM reach an agreement. This research has potential applications in education by providing efficient analysis of reflective essays and could also advance natural language processing by showcasing the effectiveness of human-LLM collaboration in specialized tasks.