Hi, I'm
The data guy — building platforms for ML/LLM at scale.
I lead a machine learning data platform team at NAVER Cloud. Our mission is to develop and operate a large-scale, fault-tolerant ML data platform where engineers and researchers collaborate to build next-generation AI systems.
Selected conference talks and panels.
We share challenges and solutions from building the AI/ML platform behind Naver’s hyperscale AI model, HyperCLOVA X. The talk includes key outcomes, future directions, and MLOps case studies from its development.
How we used AWS to algorithmically generate large-scale game worlds for Durango, leveraging services like ECS and SQS for efficient content creation and massive scalability.
An overview of building Durango's realistic ecosystem simulator using OpenCL, and how massive parallel processing reduced complex ecosystem simulations from 30 minutes to 12 seconds.
Selected work from recent roles and products.
A data management system designed for machine learning workloads, offering a Hugging Face-compatible interface for ease of adoption, along with data version control and lineage tracking.
Given a sheet of handwritten paper, generate a font that resembles the handwriting.
Ingested and aggregated operational data—from Kafka into S3 (ORC format)—for analytics, ensuring integrity with deduplication and exactly-once delivery semantics.