Writing
Notes on ML, data engineering, optimisation, and building production AI systems.
Why I reached for GPy instead of a point-estimate model when predicting building heating and cooling loads — and what the uncertainty actually bought me.
What actually happens when you point Spark at the NYC TLC taxi dataset — and why measuring execution time per transformation changed how I write pipelines.
From the closed-form solution to BFGS — what running four scipy minimizers on the same problem taught me about when gradient information actually matters.
Four queries against a mixed movie/client dataset — and the pipeline stages that made each one possible without pulling data into Python first.