nova@nova-lab: ~ — zsh

nova-lab ❯ whoami

Nova Kim — I take fuzzy asks all the way to production, solo.

AI-native Product Engineer. I run a production CRM for a US company end-to-end — payments, reporting, leaderboards — on Kotlin + Spring + AWS, where real money flows through the payment path and decisions are made on the dashboards. I turn a non-technical stakeholder’s fuzzy ask into a scoped spec with the edge cases designed in, then ship it on architecture that stays correct and stays up.

  • Payments · reporting · leaderboards in prod
  • Production CRM · solo, end-to-end
  • AI-native · guardrails + MCP
  • OSS · npx purrminal

nova-lab ❯ cat about.md

Most engineering time is lost in the gap between what a stakeholder says they want and what a system actually needs to do. That gap is where I work. A non-technical stakeholder hands me a fuzzy ask — “set me up so I can contribute to the backend and query the data,” or “we want a TV leaderboard.” I turn it into a scoped spec: explicit goals and non-goals, the constraints that surface mid-conversation, and the edge cases that decide whether it survives contact with real data — refund-then-rebill records that silently skew dashboard counts, ISO-week boundaries and timezones, null-team rows, least-privilege DB access so the safe path is also the easy one.

Then I ship it. I work AI-native — code and ship with AI — but the speed only counts because I built the guardrails that keep AI-generated code honest: ArchUnit checks that fail the build on a bad layer dependency, a spec→review loop on every PR, integration tests against a real database, health-checked GitHub Actions deploys, and backends exposed as MCP tools so agents can call them too. So the numbers a stakeholder makes decisions on stay correct, and the server stays up. Backend-first, but I own the whole line: Kotlin/Java + Spring backend, AWS infra, and an AI-assisted React frontend that ships working UI. Spec to production, solo.

nova-lab ❯ ls ~/projects

crm/   purrminal/   mentor/    # one of these hatches into a cat

crm/

A production CRM at a US company, owned end-to-end from Seoul. I take ambiguous stakeholder asks, scope them into specs with edge cases designed in, and ship payment, leaderboard, and reporting features solo — on an ArchUnit-enforced clean architecture, with integration tests on a real database and health-checked deploys, so the numbers stay correct and the server stays up.

#kotlin #spring-boot #postgres #aws #github-actions #archunit #testcontainers private repo — walkthrough on request

purrminal/

An open-source, Tamagotchi-style terminal pet — published to npm, built solo with AI leverage. Data-driven JSON for themes / evolution / i18n, hardened against corrupted state, non-TTY, and time-jump edge cases.

#typescript #ink #vitest #cli #oss run it: npx purrminal github.com/itstimi-XD/purrminal ↗

mentor/

Backend mentor at a developer bootcamp in Korea — I code-review and grade student Spring / Kotlin sprint missions, and built tooling that automates PR retrieval, retrospective verification, and grading.

#kotlin #spring #code-review #automation teaching what I ship

nova-lab ❯ cat stack.txt

  backend   →  Kotlin (primary) · Java · Spring Boot · Postgres · JPA + Flyway
  infra     →  AWS · Docker · Nginx · health-checked rolling deploys · Prometheus + Grafana
  ci/cd     →  GitHub Actions · test + ArchUnit gates · Docker build · versioned health-checked deploys
  frontend  →  React 18 · TypeScript · Vite · Tailwind · TanStack Query (AI-assisted — ships working UI)
  ai        →  AI-accelerated dev inside guardrails (ArchUnit, review loop) · lightly exploring Spring AI
  quality   →  ArchUnit · MockK + JUnit5 · Testcontainers · spec → impl → review

nova-lab ❯ contact --role 'Founding / Product Engineer'

open to US Founding / Product Engineer roles. Remote from Seoul (UTC+9), already shipping production for a US company — async/remote is solved, not theoretical. Let’s talk.