At Hanwha System's AI Tech Team, turning ad-hoc agent experiments into harness patterns the rest of the org can ship on.
Living on Claude Code, MoAI, and compound-engineering workflows. Reading every release note, re-reading the ones that matter, writing back what I learn into the team's playbook.
Embedding into real projects to cut weeks of plumbing down to days — proving the harness with code shipped, not slide decks.
Worked across Hanwha Q CELLS, Hanwha Ocean, and Hanwha Group SI — shipping data platforms on Databricks, Snowflake, Cognite CDF, and Airflow running on Azure AKS. Currently in conversations with Mistral AI about what production frontier-model work actually looks like.
click as soon as the screen turns cyan
⚠ clicking before the flash counts as a fault
Hanwha System · AI Tech Team · Seoul
Standardizing how the org ships with AI. Building harness patterns on top of Claude Code, MoAI, and compound-engineering workflows. Embedding into real projects to cut weeks of plumbing down to days. Talking to Mistral AI and external studios about what production frontier-model work actually looks like.
Hanwha System · Data Platform · Seoul
Four years building and running production-grade pipelines across the Hanwha group — Hanwha Q CELLS, Hanwha Ocean, and Hanwha Group SI. Shipped on Databricks, Snowflake, and Cognite CDF; orchestrated with Apache Airflow on Azure AKS (Kubernetes). The muscle that now informs how I think about agent systems.
Illinois Institute of Technology · Chicago
Five years in the U.S. before that — the cross-cultural muscle that makes working with international labs and English-language research feel natural.