Dear, Dear: A Coded Epistolary began with a simple archive: the text messages between myself and a person from my past. What emerged was an experiment in digital haunting—a Large Language Model (LLM) trained on our correspondence to generate words we never wrote, the conversations we might have had. Blending creative writing with machine learning, I constructed an interactive web application where users encounter these algorithmic ghosts: notes composed in the conditional tense, apologies crafted by artificial intuition, all rendered in the aesthetic of handwritten stationery. As a séance in code, this project takes the corpus of a failed romance—texts brittle with longing, coarse from unresolved fights—and runs it through an LLM like a deck of tarot cards. Rather than a prediction, the output becomes a possession. The machine spits back messages we might’ve written, conversations we almost had, alternate timelines where we were kinder, or crueller, or simply different. It’s an electronic artefact of a relationship transformed into a ouija board, where the cursor flickers between then and what if.

What began as a capstone has become a compass. I enrolled in the English Studies programme on a whim—a dart thrown blind at a board of possible futures. Two years ago, I couldn’t have told you what creative computing meant, let alone imagined it as a career. Now, that “whim” has crystallised into purpose. This project proved I can navigate the crevices between disciplines, but more importantly, it revealed how many unmapped frontiers remain. My undergraduate work ends not with mastery but with a clearer map of where mastery might lie—in those liminal spaces where code and metaphor, memory and machine, continue their uneasy dance.

Having said that, none of this would exist without Professor Collier Nogues, whose unwavering support and visionary guidance showed me how to push literary arts towards uncharted possibilities. Her mentorship transformed boundaries into bridges, and for that, I am deeply grateful.