Skip to content
View HoonStyle's full-sized avatar

Block or report HoonStyle

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
HoonStyle/README.md

Hoon Lim (임훈)

B2B Technical Program Manager — Technical Leadership · Platform Architecture · AI Agent Systems

I turn customer requirements and technical constraints into development scope, architecture decisions, and working systems. Over 15 years, I have led B2B projects across manufacturing, healthcare, and enterprise platforms.

Alongside project leadership, I build AI agents and open-source tools for code search, specification, and review workflows. I use these tools in my current work to evaluate where they help—and where they fall short.

What I build

  • legacy-spec-agent — Claude Code / Codex plugin that generates citation-backed (path:line) specifications from legacy code. Separates LLM reasoning from deterministic evidence checks and tracks citation validity as code changes. Published evaluations document both citation accuracy and gaps in coverage. MIT.
  • greplet — Local hybrid search for code and specification documents, combining vector search and BM25. Returns source locations rather than generated summaries, with workspace indexing and MCP integration for AI coding agents. MIT.
  • tierwork — Claude Code / Codex plugin for delegating review tasks and tracking their execution. Makes worker roles, review results, and reuse decisions explicit so completed work can be distinguished from proposed or skipped checks.
  • OhMoney-portfolio — Architecture and product-decision record of a 20-agent LLM orchestration system I planned, built, and operated as personal R&D. Covers LangGraph control flow, automated grading, model-routing policies, and staged pipelines with retries and fallbacks.

Where I come from

  • B2B SaaS (0→1): Co-founded and led product development for a manufacturing operations platform used across 30+ sites, with offline-first operation and eventual consistency.
  • Regulated Healthcare: Led planning and coordination for a 38-person cross-functional CGM platform TF; implementation was handled by an external development partner. The TF achieved 99.9% data integrity and passed validation against MDR/FDA requirements—not regulatory approval or launch.
  • Semiconductor / Display: Built and led MES–equipment integration work using SECS/GEM and HSMS, including 4+ years of fab-site work. Designed Core vs Plugin boundaries to reuse platform capabilities across customers.

📄 Portfolio: hoonlim.short.gy/portfolio · 💼 LinkedIn

Pinned Loading

  1. legacy-spec-agent legacy-spec-agent Public

    A Claude Code plugin that reconstructs citation-backed specs from undocumented code

    TypeScript

  2. OhMoney-portfolio OhMoney-portfolio Public

    LLM Agent-Based YouTube Shorts Automation System — Architecture Portfolio

    Python

  3. greplet greplet Public

    로컬 하이브리드 코드·PDF 검색 서버 — 레거시 여러 벌 + 현재 코드 + 사양서를 인덱싱해 AI 코딩 에이전트에 파일·심볼·줄 범위로 답한다 (LanceDB · Ollama bge-m3 · Roslyn/PdfPig · MCP/Claude Code/Codex)

    JavaScript

  4. tierwork tierwork Public

    Model-tier aware sub-agent delegation plugin for Claude Code and Codex

    Python

  5. agent-policy-radar agent-policy-radar Public

    Python