An exploration of the machine learning approach behind the Stock Advisor, examining how financial data is transformed into predictions while focusing on uncertainty estimates and feature interpretations that help separate signal from noise in the stock market.
A detailed exploration of building a modern data engineering stack: how careful tool selection and cloud-native design enable reliable data pipelines, machine learning workflows, and interactive reports with zero operational costs.
From cautious ETF investor to data engineer on a mission: I built a personal stock advisor that combines financial data with machine learning to predict long-term US stock performance relative to S&P500, running on a completely free tech stack while providing interpretable insights for investment decisions.
This post is a reply to my three-year-old post on my previous attempt to create a personal site, in which I explained why I chose R Distill (now called Quarto blog) + Firebase Hosting. I’ll outline what has changed and what hasn’t since then, which made me land on Hugo + Netlify. In the next post I’ll describe my “blogging stack”, which comprise all the tools I use from ideation to published blog post. In summary, I am happy I made the switch to Hugo and my current blogging stack works well.
An interactive dashboard designed to facilitate the exploration and understanding of a potential model for calculating a Click-Through Rate (CTR) based ranking score. It uses the Binomial-Beta conjugate prior relationship for modelling CTR uncertainty and the Beta quantile function to determine ranking order.
An interactive dashboard for visualizing article rankings on a (hypothetical) newspaper front page. Made with Svelte, running purely in the front-end. The post outlines the motivation for creating it, the ranking method used and some tooling considerations.