Jaehun Kim

Product Engineer · MVP Builder · AI/MCP

I connect frontend, backend, AI, infra, and ops until the thing works in production.

KO

I am not a one-layer engineer. If a problem needs UI, APIs, data modeling, an AI pipeline, deployment, or operations, I wire those pieces together and get the MVP in front of real users.

Most recently at Eoding, I built AI products directly on top of travel-agency ERP data. The MCP-based ChatGPT App is live on the OpenAI Apps directory; Eoding Canvas is an AI service that turns ERP product, price, and promotion data into marketing images and detail pages — combining an LLM with a multi-model image pipeline — with 4,000+ images generated across 20+ agencies.

My base is AI: an M.S. in Artificial Intelligence and three years as a deep learning engineer. Since then I have stayed close to product: ERP, web builders, payments, infra, and global launches. In crypto, I have built from an Ethereum dApp in 2018 to Bitcoin DQN trading in 2019 to an AMM implementation in 2025.

Projects

Eoding Canvas project generation screen

Eoding Canvas — An AI Service Inside the ERP

Jan 2026 – present

An AI service that reads the product, pricing, and promotion data agencies already keep in the ERP and turns it into marketing banners and product detail pages. An agency operator picks what to make, and an LLM plus a multi-model image pipeline produces it straight from existing business data — no fresh prompt-writing.

  • Built the generation request handling, job states, and result-management flow with Next.js 16 + AI SDK
  • Wired Gemini, GPT Image 2, and Grok Imagine into a multi-model image pipeline: two-stage outpainting, model fallbacks, and mask-based inpaint / erase-fill
  • Computed safe zones from 85+ production banner templates so generated subjects avoid each site's headers, forms, and text overlays
  • Added per-agency AI usage limits and generation-cost tracking so operators can see overuse and spend

20+ agencies using it4,000+ images generatedBuilt for non-technical operators

  • Next.js 16
  • TypeScript
  • AI SDK
  • Redis
Eoding app listing inside ChatGPT Apps

Eoding ChatGPT App — Travel Search Wired Through MCP

Live since Feb 2026
Open on the OpenAI Apps directory

A ChatGPT app for searching Eoding's live ERP travel inventory. The LLM calls MCP tools over 15,000+ products and 6,000+ cities, then renders results through in-chat widgets.

  • Built the OpenAI Apps SDK + React widgets, MCP server, and TypeScript/Express search API
  • Mapped travel-domain filters like destination, departure date, and trip duration into safe tool parameters
  • Listed publicly on the OpenAI Apps directory, so the result is externally verifiable

Live on the OpenAI Apps directoryMCP/API/widgets wired end to endBacked by live ERP data

  • MCP
  • OpenAI Apps SDK
  • React
  • TypeScript
  • Express
miniAMM live demo UI

miniAMM — Uniswap V2 AMM Implementation

Sep 2025 · Hell Month Cohort 1

A hands-on Uniswap V2-style AMM built during Joel Mun's Hell Month Cohort 1. I connected the contracts, local chain, explorer, and frontend into one working flow.

  • Wrote swap and add/remove-liquidity contracts in Solidity and tested them with Foundry
  • Composed a private geth chain and Blockscout explorer with Docker Compose
  • Built wallet connection and UI with Next.js 15, wagmi/viem, ethers v6, RainbowKit, and TypeChain

I have touched crypto by building: 2018 dApp, 2019 trading agent, 2025 AMM

  • Solidity
  • Foundry
  • geth
  • Blockscout
  • Next.js 15
  • wagmi/viem
Live demo miniAMM-front Hell Month Cohort 1

Experience

  1. Jul 2023 — present

    Product Engineer Eoding

    • Own new products that wire ERP data to LLMs and image models: ChatGPT App, Eoding Canvas, and related AI workflows
    • Led the global launch — localization, KRW–USD multi-currency, multi-region scalability
    • Expanded Terraform · GitHub Actions CI/CD including production rollbacks
  2. Sep 2021 — Jun 2023

    Full-Stack Engineer Eoding

    • Built the Next.js Web Builder used by 200+ paying agencies so non-technical staff can edit their own sites
    • Implemented four-PG payment integration, shared auth, promotion rules, and realtime reservation/sales dashboards
  3. Dec 2019 — Aug 2021

    Software Engineer Eoding

    • Co-developed core ERP modules for product/reservation workflow automation and realtime website sync
    • Led the modular Web Builder (React/Next.js)
  4. Jan 2017 — Dec 2019

    Deep Learning Engineer Finfra

    • Plant-growth computer vision for a KIST smart-farm project — detection & CNN measurement
    • Autoencoder anomaly detection over 300GB+ logs for KT DS — distributed TensorFlow · Hadoop
  5. Sep 2016 — Jan 2017

    Cloud Researcher (DL · Big Data) KSIC

    • Distributed GPU TensorFlow setup and Hadoop cluster environment

Early Work (2017–2019)

Early AI, blockchain, and computer-vision projects from 2017–2019.

Deep Trading Agent — Original write-up EtherMembership — Original write-up Fish Detection — Original write-up Plant Growth Analysis — Original write-up

Skills

Product
MVP building · problem framing · non-technical workflows · production ops
AI/LLM
MCP · OpenAI Apps SDK · AI SDK · LLM/API integration · multi-model image pipelines
Full-Stack
React · Next.js · TypeScript · Node.js · Kotlin · Django · Python
Blockchain
Solidity · Foundry · wagmi/viem · ethers
Infra
GitHub Actions · Terraform · AWS · Docker · Redis

Education

  • University of Southampton — M.S., Artificial Intelligence (2014–2015)
  • Sangmyung University — B.S., Computer Science (2007–2014)