2020.05 – Present · NAVER data organization — Place platform (org leader since 2021.04)

Leading a sub-10-person data engineering org in NAVER's data organization — transformed manual data request handling into a contract-based self-service platform, and now building a dev culture where AI agents work alongside the team.

AI Engineering & Dev Workflow Automation

• Led LLM-agent (Claude Code) based team automation — built and operate an AI code review bot, automated Jira issue triage/operations, and an internal AI assistant (Slack bot) on Kubernetes. Automated triage cut median external data-request lead time from ~15 days to 3 (measured before/after).
• Designed AI contribution measurement infrastructure — deterministic tracking signatures aggregated via LLM judgment: 400+ merged AI-authored PRs and 1,200+ AI-driven tasks measured within six months of adoption (as of Jul 2026).
• Standardized repetitive work (data modeling → pipeline contract → access requests) into reusable AI skills/plugins shared by the whole team.

Data Lakehouse

• Designed and operate an Iceberg·Spark·Kafka·Airflow data lakehouse — a dedicated Hadoop cluster (double-digit node count) with a triple-digit-TB snapshot layer, realtime CDC from 13+ service sources, and active/standby disaster recovery.
• Led two large-scale zero-downtime migrations — a full cloud-platform transition (Spark on YARN→K8s, 66% legacy cluster reduction) and a triple-digit-TB IDC move (dual-write / fade-in-out, ~25% hardware savings) — with no service interruption.
• Root-caused a defect in Iceberg auto-compaction, cutting rewrite time by ~98% and improving data freshness by up to an hour.

dbt-based Semantic Layer & Dimensional Modeling (Place BI · 100+ Metrics · NBoost Award)

• Built and delivered Place BI on top of the dbt Semantic Layer, serving 100+ core metrics as a Single Source of Truth (SSOT) — collaborating closely with the DA (Data Analytics) organization and leading the dev team to standardize company-wide metrics, recognized with the 'NBoost Award' (2026.08.16).
• Established dbt-based dimensional modeling conventions (fact/dimension/serving layers) and built a static validation tool that auto-detects violations — model quality guaranteed by the system, not by people.
• Designed and configured dbt Semantic Layers — built an automated verification harness linking Canonical Metrics with Binding Metrics, completing enterprise-wide data integration and standardizing 100+ core metrics.
• Standardized metrics into KPI dashboards via MetricFlow, eliminating cross-team data inconsistencies and continuously verifying KPI parity across heterogeneous systems.
• Built an automated parity-verification harness that cross-checks KPI numbers produced by different systems, securing metric trustworthiness.

Data-Contract-Based Self-Service ETL

• Designed a self-service data platform where schema definition (Protobuf), pipeline creation, and data subscription/permissions are all handled by a single contract-file PR — operating 500+ data models, 30+ pipeline contracts, and 160+ dbt models.
• Supplies data to AI and analytics workloads across the company through 12 data products (DaaS).

Data Governance & Data Mesh

• Led data governance for the Place data domain and headed a Data Mesh transition TF, designing a domain-oriented data ownership structure.

Realtime Abuse Detection

• Advanced a realtime-pipeline-based abuse detection system, raising service trustworthiness. Tech blog (KR)

2018.05 – 2020.05

Learning new technologies and concepts, then turning them back into user-facing services — made blockchain, an unproven technology, into a wallet service for a global messenger's users and production-grade infrastructure.

• Developed the wallet service managing crypto assets for LINE messenger users — connecting new-technology learning to a real user-facing service.
• Solved database sharding differently from the RDB approach with an HBase-based design — widening my spectrum of large-scale data processing techniques.
• Researched and developed blockchain transaction performance improvements and interchain systems — filed 3 related patent applications.
• Built KMS/HSM-based secure blockchain network provisioning and Docker/Kubernetes node orchestration. Tech blog

2016.10 – 2018.05

A move I chose to close the data gap I felt building settlement systems — working across server development and data engineering on a global messenger's ad platform, recognized as best engineer for it.

• Developed a high-performance ad event collection system handling 2B daily requests / 4M TPS (Kafka, Avro, ELK, Hive).
• Built the servers delivering ads through LINE Official Account channels and the admin permission-management module, and designed user-demography-based ad targeting (Spring, Scala, Akka-HTTP).
• Collaborated with the Finland office on an ad creative renewal — experiencing the ad product beyond server engineering.
• Received the Best Engineer Award (2017) for outstanding technical contribution and business impact.

2011.01 – 2016.10

Community, search, music streaming — the years I built global services users spend their time in, from the ground up, learning both large-scale fundamentals and the joy of building products.

LINE MUSIC — Streaming Service & Royalty Settlement

• Designed and built the service backend end to end, from the search engine to the streaming APIs — the experience of making a service users spend part of their day in (Spring, MySQL, Elasticsearch).
• Built the music search server on Elasticsearch from scratch — designing tokenizers and analyzers and applying n-gram/bigram matching for partial-match search on track and artist names, internalizing how search engines work (the foundation that later led to Jest open-source contributions and the ES-based settlement system).
• Designed the playback-time-based royalty formula and settlement system on Elasticsearch/NoSQL — producing time-critical monthly settlement reports for 60+ music labels, without Hadoop.
• Built the IAP/IAB in-app billing system (Spring Batch).
• This settlement system became a career inflection point — realizing large-scale data processing (Hadoop) was my next gap, I chose to move to the ads team to close it.

LINE Messenger Note — a Community Service Inside the Messenger

• Built a high-volume community board service running inside a global messenger — handling messenger-scale traffic on sharded MySQL, and experiencing a new kind of product: services on top of a messenger.

Global Community & Search

• Developed Naver Japan Cafe (community service for Japan) — the start of hands-on Japan-facing service development.
• Built LINE Thailand shopping search (Elasticsearch).

2008.07 – 2010.12 (internship 2008.03 – 06)

From intern to Rookie Award in year one, Excellent Employee Award the next — proving growth speed from the start.

• Led game logic and backend server development for 'Hoyfarm', the predecessor of the hit social game 'Rule the Sky'.
• Designed and implemented a search service for game community sites end to end, from the web crawler to the search engine — digging into Lucene to deliver not just per-site search but unified search across sites. The starting point of a search journey that later led to LINE MUSIC's Elasticsearch search and Jest open-source contributions.
• Converted to full-time after the 2008 internship; won the Rookie Award and Excellent Employee Award in consecutive years.