Minsoo Kim · Seoul, KR

I turn bare metal into data platforms.

Data platform engineer, ten years in — started in iOS, moved through ML, and settled where I'm most useful: the layer everything else runs on. Currently building an MLOps platform on bare-metal Kubernetes, end to end, solo.

github linkedin email

Career, as a medallion architecture

Data engineers refine raw events into silver, then gold. A career works the same way — each layer feeds the next.

Bronze raw ingestion — learning how software is actually shipped

2018

Finple — DevOps & hybrid apps

Maintained iOS/Android hybrid apps, built Node.js scrapers.

2019–21

Eazel — iOS engineer

Led the ReactorKit reactive architecture and API-client design; introduced CI/CD.

Silver refinement — from app code to data & ML systems

2021–22

Eazel — MLOps engineer

NLP modeling to validation end-to-end; social-network-analysis dashboard that helped land Series A funding.

2022–24

Eazel — team lead, AI Labs

Art-market ETL on AWS EMR/Airflow, embedding-based dedup with vector DBs, LangChain RAG over PostgreSQL & Neo4j.

Gold aggregation — running the platform layer itself

2025

Cellkey — data engineer

StarRocks cluster on EC2: analytical queries from 7s to 1.4s (−80%).

2025–

HUINNO — data platform engineer

In-house MLOps platform on bare-metal Kubernetes, built from scratch with Pulumi: Iceberg lakehouse on tiered storage, HA Airflow + dbt, KubeRay across GPU nodes, MLflow & DataHub. One person, whole stack.

Runs in production today

The tools I operate daily, not a keyword cloud.

Kubernetes (bare-metal) Pulumi Apache Iceberg Airflow 3 dbt DuckDB Ray MLflow DataHub SeaweedFS PostgreSQL Python Swift