Blog
Thoughts on practical data science, ML engineering, career building, and lessons from production systems.
In a 16-hour Fundamentals of Machine Learning course I trained a chest X-ray classifier in Teachable Machine in front of the students, gave it an interface with Google AI Studio, and later rebuilt it with Claude Code into a browser app that runs any Teachable Machine model.
Why the Olaia Labs landing page needed more than a contact form, and how a five-step readiness questionnaire became a lead tool that gives value before it asks for an email.
For a 15-hour introductory AI course for accountants I needed a realistic example, not a toy. Here is the n8n workflow that reads invoices from Gmail, including scanned ones, extracts the data with an LLM, files the PDF in the right client folder, and chases approvals by itself.
How a K-Means example I built for a data analysis course in November 2024 turned, one prototype at a time, into SegmentAI Analyst: a web app that segments customers and writes the marketing strategy for each segment.
A practical guide to deploying interactive ML demos using Hugging Face Spaces, from setup to sharing your models with the world.
Your GitHub profile isn't enough. Here's why a personal website is the highest-leverage investment you can make for your data science career.
Lessons learned from production ML systems: how to design pipelines that handle failures gracefully, log meaningfully, and alert before things go wrong.