I study AI and Data Engineering at the Technical University of Denmark and am currently on exchange at UC Santa Cruz. I am interested in machine learning systems, data pipelines, and careful evaluation of models.
DTU group project fine-tuning a compact BERT model to classify text readability. The workflow connects data preparation and versioning with training, evaluation, FastAPI serving, Docker, and cloud deployment. My contributions included CI and cloud-build configuration, with DVC data/model handling in the build workflow.
DTU group project exploring model training on an Arduino Nano 33 BLE Sense Rev2 and classification of peripheral nerve signals. The group demonstrated on-device MNIST training; the nerve-signal work remained limited by preprocessing. My contributions included on-device experiments, signal preprocessing, and adaptation demos.
Hackathon team project helping educators review AI-suggested literacy errors in handwritten work. I contributed handwriting review with reversible decisions, page orientation with aligned overlays, error-trend charts, and a resumable dataset downloader.
DTU group project training a CNN to classify chest X-rays, with experiments in image preprocessing, model training, and evaluation.
Independent Chrome extension that summarizes a YouTube video's available captions before you open it.
Python, PyTorch, scikit-learn, pandas, FastAPI, Docker, DVC, and GitHub Actions.