Skip to content
View jungrok5's full-sized avatar

Block or report jungrok5

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
jungrok5/README.md

Jeongrok Oh · 오정록

Technical Director · MMO Server Architect — 20 years building game servers (since 2007), at NCSOFT since 2016.

I build MMO server frameworks from scratch — networking, lock-free concurrency, navigation, server-side physics and virtual-actor distributed systems — and I optimize across the whole stack: server, client and database. I prefer leading technology hands-on to leading it through management.

📄 Resume (English) · 이력서 (한국어 · 전문) · 3D interactive 💼 LinkedIn · ✉️ jungrok5@gmail.com


Selected results — all measured

13M packets/sec 1,000-player single-view combat, scaled to 2,000 vs 2,000 · traffic −36% — NCDP 2019 Grand Prize
50M props full-tick in one space with 2,000 concurrent builders (UE5)
20,967 → 503 draw calls, with FPS 16.5 → 47.4 (instancing + physics chunking)
240K req/sec gRPC load-test ceiling · 50K TPS on EntityFramework
1,000+ cursors synchronized in a single browser space, conflict-free
2,000 → 8,000 stable concurrent users, 3-month optimization assignment

Personal projects

Experiments in how far each domain can be pushed with AI and models as tools.

Project What it is Notable
Hellfarm 🔒 Hack-and-slash action game (Godot · C#) 128M entities simulated concurrently; dense-horde O(n²) cut from 2,500 ms to 38 ms per frame
ZeroLlama 🔒 Agentic coding client for local models (TypeScript · llama.cpp) Runs a 35B-class model on 8 GB VRAM by offloading MoE experts to CPU RAM
ondevice-imagen On-device image generation (Kotlin · ONNX Runtime) Stable Diffusion inference pipeline implemented natively for Android
one-scroll-bible The Bible as one scroll, 200+ languages Live at one-scroll-bible.com
thesauros Equity buy-timing analysis and alerting Production: collection → analysis → backtest → alerts

Stack

C++ 11–20 C# TypeScript Python Lua GDScript Kotlin Orleans gRPC/TCP IOCP ASP.NET Node.js sharding MSSQL Redis/Garnet MongoDB EntityFramework Unreal Engine 3/4/5 Unity3D Godot Recast/Detour voxel navmesh MCP (own implementation) llama.cpp ONNX Runtime

Pinned Loading

  1. mmo-lag-anatomy mmo-lag-anatomy Public

    Game Lag White Paper: an interactive encyclopedia of 228 online game lag causes across 13 layers, from client to database, each with symptoms, graph shape, how to confirm, owning team and sources. …

    HTML

  2. zerollama-dashboard zerollama-dashboard Public

    Single-HTML monitoring dashboard for llama.cpp — per-slot sampling, plain-language tuning hints, router-mode aware, with built-in chat. Zero deps.

    HTML 30 3

  3. ondevice-imagen ondevice-imagen Public

    On-device Stable Diffusion (1.5 / Turbo / XL) on Android — ONNX Runtime + Godot + Kotlin. Part of an AI research notebook series

    Python 1

  4. one-scroll-bible one-scroll-bible Public

    📖 The whole Bible as one scrollable mobile page — Creation → Cross → Restoration, leading to the gospel. In many languages. · 한눈에 보는 성경 이야기

    HTML 7