I've built an end-to-end Python ETL pipeline at Digital Research Lab for time-series sensor data from hundreds of oil and gas wells, translating operational challenges into measurable analytical requirements and modelling approaches.
I developed statistical feature-engineering and detection logic for sustained pressure behaviour, trends, and anomalies, then delivered KPIs and actionable findings through a SQL-backed dashboard for technical and business stakeholders.
For my M.Sc. thesis, I'm improving deterministic policy-gradient meta-learning for combinatorial graph optimization. I built a reproducible JAX and MLflow experimentation pipeline that matched baseline solution quality while training 2.7× faster on large instances.
I've also supported graduate optimization teaching and automotive image-data quality checking, and I bring hands-on experience in Python, SQL, machine learning, deep learning, time-series analysis, and data visualization.
