Hi, I’m Bathila Jayasiri
I’m a software engineering intern with a background in data science and astrophysics. I’m exploring the space where data, machine learning and software meet.
Client project
A pipeline that translates e-learning videos end to end: slide text, spoken audio and burned-in subtitles. I built the computer vision and speech stages, integrated machine translation and neural text-to-speech, and tested it across multiple languages.
Python / OpenCV / EasyOCR / Whisper / FastAPI
MSc dissertation
I built a modelling dataset from synthetic UK banking transactions in the FCA Digital Sandbox, then compared five approaches to fraud detection. LightGBM reached 97.8% recall and 84.9% precision at a 4% alert volume. I used SHAP to examine which features influenced the model’s predictions.
Python / SQL / LightGBM / scikit-learn / SHAP
Machine learning project
I predicted daily passenger flow across the TfL network from 2019 to 2024. I combined station footfall with weather, bank holidays and a COVID-19 restriction score, then trained gradient boosting models and tuned XGBoost with a cross-validated grid search.
Python / pandas / scikit-learn / XGBoost
Imperial College London
Eight unknown functions, with one query per function each week. I used Gaussian processes and Expected Improvement to decide where to sample next. Over ten rounds, I improved the starting best result on six of the eight functions.
Python / scikit-learn / NumPy / SciPy
BSc dissertation
I fitted WASP-121 b transit data across 24 wavelength channels using BATMAN and MCMC, then constructed a transmission spectrum and compared the radius-ratio estimates with published results.
Python / BATMAN / MCMC / Matplotlib
Skaleminds
University of Birmingham
Queen Mary Students’ Union
Imperial College London
Professional Certificate
University of Birmingham
MSc
Queen Mary University of London
BSc (Hons)
For opportunities or a conversation about my work.
[email protected]