Artificial intelligence postgraduate personal statement example

Personal statement example

On Thursday evenings I climb forty-two steps to the ringing chamber of my parish church and ring the fourth of six bells. Change ringing has no tunes. Each row is a permutation of the bells, and a method such as Plain Bob Minor is a rule for moving from one row to the next so that no row repeats. When I started, I memorised blue lines on paper. After a year I realised I was holding a small search problem in my head and pruning it as I went. That sense of following a rule-governed process, and noticing when it goes wrong, is a large part of why I want to study artificial intelligence at postgraduate level.

My degree in Electronic Engineering gave me a solid grounding in signals, linear algebra, probability and programming in C and Python. The module I enjoyed most covered digital signal processing, because it showed me how much structure can be extracted from data that looks like noise. For my final-year project I used a vibration rig in the department teaching lab to collect accelerometer data from a small motor fitted with healthy and deliberately damaged bearings. I compared two approaches to classifying faults: a random forest trained on features I calculated myself, such as RMS amplitude, kurtosis and energy in frequency bands around the bearing's characteristic frequencies, and a compact one-dimensional convolutional network trained on the raw windows.

Both performed well when tested at the motor speed used for training. The more useful result came when I tested at a different speed. The convolutional network dropped noticeably, while the hand-built features held up better because I had normalised the frequency bands to shaft speed. I would not claim this as a general finding from one rig and a few hours of recordings, but it made the idea of distribution shift concrete for me. It also left me wanting to understand properly why learned representations generalise in some situations and not others, rather than treating a training curve as the whole story.

Since graduating I have worked as a test technician at an appliance repair centre. Most of my day involves safety-testing returned washing machines and dishwashers, logging faults and passing units to repair engineers. I have become quick and methodical, and I have learned to write fault notes that someone else can act on without asking me what I meant. I also built a simple Python script that groups our free-text fault descriptions by keyword so my supervisor can see which models return most often. It is basic, but it showed me how messy real records are, and how much judgement sits behind any label.

Outside work I have been reading Russell and Norvig's Artificial Intelligence: A Modern Approach alongside online lectures, concentrating on the chapters on search, probabilistic reasoning and learning. Working through A* search after years of ringing methods was satisfying, because heuristics finally had a formal shape. I have also started implementing small examples myself rather than only reading about them.

My other regular commitment is helping my grandmother with letters and online forms, often explaining official decisions that have been written badly. It has made me care about systems that people can question and understand, which I expect to be an increasingly important part of how AI is built and deployed.

I am applying for a Masters because I want depth in machine learning theory, reasoning under uncertainty and robust evaluation, and the chance to work on a substantial project with proper supervision. I bring an engineer's habits of measurement and testing, steady experience of careful practical work, and a genuine curiosity about how rule-following systems behave when the world does not match their assumptions.