<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Yoonmee Hwang</title><link>https://devotto.github.io/</link><atom:link href="https://devotto.github.io/index.xml" rel="self" type="application/rss+xml"/><description>Yoonmee Hwang</description><generator>HugoBlox Kit (https://hugoblox.com)</generator><language>en-us</language><lastBuildDate>Wed, 01 Jul 2026 00:00:00 +0000</lastBuildDate><image><url>https://devotto.github.io/media/icon_hu_1c0e9cb08cfb822a.png</url><title>Yoonmee Hwang</title><link>https://devotto.github.io/</link></image><item><title>ML Experimentation Platform for Time-Series Anomaly Detection</title><link>https://devotto.github.io/projects/ml-experimentation/</link><pubDate>Mon, 11 May 2026 00:00:00 +0000</pubDate><guid>https://devotto.github.io/projects/ml-experimentation/</guid><description>&lt;p&gt;Independent study. Reproducible PyTorch experiment framework evaluating 5 model architectures across 200+ time-series datasets on a Slurm HPC cluster. Fail-fast screening cut a 17 GPU-hour benchmark to 1.3 GPU-hours, with 600+ runs tracked in Weights &amp;amp; Biases.&lt;/p&gt;</description></item><item><title>Movie Recommender System in Production</title><link>https://devotto.github.io/projects/movie-recommender/</link><pubDate>Sun, 01 Feb 2026 00:00:00 +0000</pubDate><guid>https://devotto.github.io/projects/movie-recommender/</guid><description>&lt;p&gt;Infrastructure and MLOps lead in a 5-person CMU team (17-645). Ran a movie recommender for a simulated 1M-user service on k3s with Helm, GitHub Actions CI/CD, and Prometheus, Grafana, and Loki monitoring. Cut timeout rate to 3.3% and cold-start latency to under 100ms (dev-measured).&lt;/p&gt;</description></item><item><title>Financial Integration Service (Troutwood)</title><link>https://devotto.github.io/projects/troutwood/</link><pubDate>Fri, 01 Aug 2025 00:00:00 +0000</pubDate><guid>https://devotto.github.io/projects/troutwood/</guid><description>&lt;p&gt;Serverless payroll-data ingestion platform on AWS (Lambda, SQS, DynamoDB, CDK), built as a CMU MSE Studio project for
. Designed the end-to-end CI/CD platform on GitHub Actions: 14 of 17 workflows, gated deploys, per-branch ephemeral stacks, GitHub OIDC auth, and LocalStack integration tests.&lt;/p&gt;</description></item></channel></rss>