Junkmail AI Cleaner
A public Outlook cleanup tool that uses fast rules for known spam and a local language model for uncertain messages. Its documented safeguards focus on keeping legitimate mail and stopping safely when dependencies fail.
Melbourne, Australia · Engineering
Make complex systems clearer to run.
I’m Song Jin. My published work covers Kubernetes, cloud infrastructure, and delivery automation. I’m also exploring practical ways to improve knowledge-heavy workflows with AI.
Selected work
Work you can inspect
Specific problems and tradeoffs tell you more than a list of tools.
↗
Detecting Kubernetes API latency
An investigation into alerting on sudden API server latency. The published note compares a slow summary metric with a histogram-based Prometheus query and describes how the scenario was reproduced.
The AI tool is a public personal project. The Kubernetes note is an archived investigation. Neither is presented as a client case study.
About
Engineering with an eye for the whole workflow.
I’ve written about Kubernetes operations, monitoring, and CI/CD from hands-on engineering work. The archive shows how I reasoned through implementation details, including approaches that did not work first time.
For new work, I’m interested in the point where platform engineering and useful AI tools meet: making complex tasks easier to understand, repeat, and improve.
Read the engineering archiveFor teams and organisations
Have a workflow worth improving?
I’m exploring practical AI-assisted workflow improvement: starting with the work people repeat, the knowledge they need, and the checks that make automation trustworthy.
The engineering examples here show how I investigate problems and explain tradeoffs. Explore the public work to see the approach in detail.
View my public work