Software engineer · Data & automation
I'm KT, a software engineer with ten years of experience building systems that run reliably without anyone babysitting them.
My work centers on production machine learning and data automation, and I like replacing manual, spreadsheet-driven processes with tools that just work.
I hold a Master's in Data Science from UC Berkeley, where my capstone, an end-to-end system for detecting oil spills in satellite imagery, won the Hal R. Varian Capstone Award. Outside of work I build practical tools, like a meal planning app that turns recipes and weekly menus into grocery lists.
A meal planner for the whole week
The Menu stores recipes, plans daily menus, tracks the pantry and builds the weekly grocery list. It is plain HTML, CSS and JavaScript, with Supabase handling accounts and the database, deployed on Vercel.
My family uses it every week to plan our meals, and a few friends have made accounts too.
Automated oil spill detection from satellite radar imagery
OceanWatch is an end-to-end system that monitors chosen ocean areas for oil spills and alerts users when one is detected. It was my capstone project in UC Berkeley's Master of Data Science program, built with my team, and it won the Hal R. Varian Capstone Award.
The motivation: after a tanker capsized off Trinidad and Tobago, it took three days for anyone to notice, and the oil spread as far as 100 km to neighboring islands by then. Faster detection means faster response.
We trained and compared several image segmentation models on a public dataset of labeled satellite radar (SAR) images to find spills. The pipeline runs on AWS (Lambda, EC2 and S3), automatically monitoring user-defined areas of interest and sending notifications.
I'm currently exploring web analytics and AI-assisted workflows.