Places I've Worked

I joined as a strategic technical advisor reporting directly to the CEO Office. Korea’s largest cloud MSP — about 2,000 employees, multi-cloud (AWS · GCP · Azure) — with a seat at most company-wide decisions.

Technical Advisory

• Reviewed ongoing projects for technical gaps; validated requirements across the organization
• Cross-functional advisory across multi-CSP environments

’by Megazone’ Product Guidelines

• Established SDLC (Software Development Life Cycle) guidelines
• Codified engineering management practices

I started as an advisor and moved into the CTO and director of the research institute role. I ran an R&D organization of about 20 AI developers and software engineers.

Core Technology

• Deep Temporal Neural Networks–based Explainable AI (XAI)
• Commercialized via technology transfer from UNIST and KAIST
• Product: INFINITE OPTIMAL SERIES™ — industrial process optimization solution

Industries

• Steel · cement · glass · petrochemical · power · transportation and logistics

Engineering Operations

• Introduced state-of-the-art procedures — improved deployment reliability and model trust
• Managed 20+ AI/SW engineers

I wanted to explore how LLMs can fit into a classroom. I started Aisle Street with a co-founder and we’re building exemplar.ai.

Product

• exemplar.ai — a closed chatting service for teachers and students
• RAG-based, supports the use of AI inside lessons
• Teacher–student chat with content governance over course material

Role

• Co-Founder, tech lead — backend architecture and full-stack development

Tech Stack

• React, AWS, Pinecone, RAG, LLM integration

Currently operated as a non-profit / portfolio project.

I wanted to translate the field experience from Google Search, Play Store, and AI work into a classroom. About 80 Digital Finance MBA and Information Management students each semester — a 3-credit, 15–16-week in-person course taught for three consecutive years (2023–2025).

Course: AI and Recommendation Systems

• Audience: Digital Finance MBA + Information Management + Management Engineering (~80 per semester including auditors)
• Format: in-person lectures + coding labs + student presentations

Curriculum

• Recommendation systems for 9 weeks (collaborative filtering and other algorithms, MovieLens dataset)
• AI fundamentals + State of AI + weekly news follow-ups
• Hands-on: OpenAI API usage, prompt engineering
• Special sessions: Google Play Store case study (from my own experience), search quality evaluation
• Final: ~47 student presentations, 15 minutes each

Approach

• Theory + coding labs + current AI trends as one curriculum
• Bringing a product perspective into the academic setting

I publish lecture notes on Brunch each semester. Material that feels too good to use only in the classroom.

I wanted to bring thirteen years at Google back to a Korean startup. I joined Banksalad as VP of Engineering in a two-person engineering leadership setup with the CTO, taking on roughly half of the 80-person tech organization.

Organization & Hiring

• Direct management of 40+ engineers (SWE · DS · DA · DevOps · EM) — half of the tech org
• Hired ~20 engineers over 1.5 years. Led senior interviews across Data, Backend, and Frontend (Web)
• Owned the Web chapter directly — alongside Backend, Data, and DevOps chapters in a full-stack organization
• Ran biweekly 1:1s consistently across 40+ reports. Led performance reviews and promotion decisions

75% Cloud Cost Reduction (6 months)

• AWS environment. Phase 1 right-sizing cut costs by half; Phase 2 optimization cut the remainder by half — about 75% in total
• Set direction with the team and ran analysis together; personally proposed and executed the Mydata optimization
• Mydata methods: removed unnecessary API calls, optimized Redis reads, introduced delayed refresh, allocated resources by user tier

Mydata Service

• Mydata: a Korea-regulated service that lets users consent to having their financial data collected
• 258+ connected financial institutions (banks · cards · securities · insurance · pay money · installment finance)
• 5x increase in average institutions connected per user, ~2s real-time asset updates (20x improvement), ~30s end-to-end connection
• Architecture: Go + gRPC + Kubernetes + AWS + Redis microservices

Management

• Introduced experiment and project review processes — defining impact upfront resolved cross-team conflicts
• Shifted prioritization and resource allocation toward impact-based criteria

When the 2023 funding winter forced the company to cut the team in half, I — among the executives based in the U.S. — agreed to step out as well. A practical decision in the startup environment.

After many years at Google, I wanted to step outside for a while. Samsung Semiconductor’s storage team was looking for application-side insight, and it dovetailed naturally with my master’s background in Computer Architecture.

Leadership Training

• Trained Samsung Semiconductor leadership on Google-scale service requirements
• Analyzed indexing and search execution patterns from memory and storage perspectives
• Captured the hardware requirements behind preparing and running large-scale services

Data Center Business Analysis

• Filled the technical gap in the storage team’s planning and requirement work
• Translated application-tier workload patterns into concrete memory and storage requirements
• Validated whether bespoke rack configurations supplied to data centers met those requirements

A short pivot from software toward hardware. Bringing Google-scale service experience and graduate-school Computer Architecture into the same room turned out to be an interesting moment.

TLM — Core Systems & Experiences (2020.12 – 2021.10)

• Cloud Logging team, planet-scale debug log infrastructure for Google engineers

Staff SWE — Google Lens (2018.09 – 2019.12)

• Korean on-device OCR for Pixel 4
• Lens search integration analysis pipeline

Staff SWE — Nest Store (2017.12 – 2018.09)

• Payment service for store.nest.com (Hybris SAP, AWS + GCP)
• Affirm BNPL integration, 100% solo. Google’s first product with subscription sales

TLM / Product Owner — Play Store Korea & Japan (2016.03 – 2017.12)

• Google’s first region-specific consumer product, 8 SWEs. Korea & Japan = 70% of global Play Store revenue
• Locality-aware search & recommendation, pre-registration campaigns
• Prevented major game title migration to competing stores

Head of Engineering / Product Owner — Search App Indexing (2012.09 – 2016.03)

• Initiated and led App Indexing from a single demo to global scale
• Built 50-person org (4 teams). New paradigm: crawling native app content via Android simulator
• Featured in Google earnings calls, Android K release, Google IO 2015 main session speaker
• ~90% of team (45 of 50 people) promoted during 3-year project

TLM — Search App Search (2011.09 – 2012.08)

• Launched App discovery in Google Search, 5 SWEs

TLM — Search Korean Mobile (2010.09 – 2011.08)

• First manager role, 8 SWEs, 10+ mobile features in 1 year

Senior SWE — Search Rich Snippets (2009.09 – 2010.08)

• Movie vertical with Rich Snippets — contributed to schema.org standard definition
• Foundation for what became Google’s Knowledge Panel

Senior SWE — Search Korea (2007.07 – 2009.08)

• Early Seoul office member (~10 people). google.co.kr search result page redesign
• Introduced result groups architecture. Precursor to Universal Search and Knowledge Panel

The last role before Google. A research-institute setting where tools and automation were the work itself.

• Eclipse-based device driver source code generator

Set-top-box firmware → Linux kernel → middleware → Symbian porting → device driver generation. Eleven years from the layer closest to hardware up to the whole system. By the end of this chapter I had become an engineer who knew the stack from the bottom up.

A step deeper into mobile operating systems.

• Symbian / Series 60 base-porting
• Notable product: Samsung Z600

A close look at how Korean handset platform code reaches global standards.

Seven or eight years before Android, a small group set out to build a smartphone on Linux.

• Linux-based smartphone (ALL-about) development
• Owned the kernel and middleware layers above the hardware
• Contributed to the TynuxBox developer tooling
• Built Samsung handset prototypes

When I left mobile for search later, I thought I was moving on. Years later App Indexing, Play Store, and Pixel 4 OCR brought me back — this was the first foothold of that arc.

The first full-time job after finishing my master’s. Where I first learned how small code could get inside an embedded system.

• Software for Humax set-top boxes
• ECMAScript parsing module for a browser running on RTOS (pSOS)
• Wrote a JavaScript interpreter from scratch under tight resource constraints

A close-up view of how JavaScript actually breathes inside a browser.

The first place I worked, part-time, during my student years.

• Developed and shipped packaged applications including CGI Builder (a web–database integration tool)

Some of the first code I wrote, at the early days of the Korean web in the 1990s.

M.S. — Computer Engineering (Computer Architecture) (1997.3 – 1999.2)

• Deep grounding in hardware, firmware, and embedded systems
• Foundation that later connected to on-device ML (Google Lens, Pixel 4) and semiconductor advisory (Samsung Semiconductor)

B.S. — Computer Engineering (1993.3 – 1997.2)