HARUMI

About

Harumi Morikawa

Platform / Backend Engineer

I am a platform-minded software engineer who learns by operating, not by watching tutorials. I run a two-node Kubernetes cluster end to end — GitLab, ArgoCD, Camunda, and my own applications — and I treat every incident on it as course material that I write up afterwards. What you see on this site is not a portfolio of demos; it is the running system itself.

Location
Based in Japan
Languages
中文 · 日本語 · English
Focus
Platform engineering · Backend · Workflow systems
Links
GitLab · LinkedIn

How I work

  • Verify against the source of truth

    A debug pod is not the node. ArgoCD “Synced” is not a deploy. I check the real thing — Envoy stats, the host’s own systemd, the actual tag-bump commit — before calling anything done.

  • Remove, don’t layer

    When a speculative fix turns out to be wrong, I take it out and write down why — instead of stacking a second fix on top of it.

  • Ship, listen, revert honestly

    I shipped a front-panel display hack, heard “not legible at a glance”, and reverted it rather than defending it. The lesson stayed; the bad version did not.

Capabilities

What I work with

OPERATE

Kubernetes · GitOps · Gateway API · Linux

BUILD

Spring Boot · Kafka · PostgreSQL · Next.js

INVESTIGATE

Envoy · Distributed systems · Incident analysis

Show the full technical list

Platform & operations

  • Kubernetes
  • ArgoCD / GitOps
  • GitLab CI
  • Gateway API / Envoy
  • Cloudflare Zero Trust
  • systemd / Linux

Backend

  • Spring Boot
  • Kafka
  • PostgreSQL
  • Redis
  • Camunda 8 / Zeebe
  • Keycloak / OIDC
  • WebSphere / Jython

Frontend

  • Next.js / React
  • TypeScript
  • PWA / Web Push
  • i18n (EN / JA)
  • CSS motion design

Ways of working

  • Incident write-ups
  • Two-stage GitOps
  • Evidence over guesswork

CURRENT / 2026

Current focus

I am currently extending my platform and backend experience into AI-assisted engineering systems — operational search, knowledge retrieval and workflow automation.

Platform reliability

AI-assisted document processing

Operational writing

GitLab activity

Daily activity from the self-hosted GitLab — counts only, no repo names or messages.

Loading…

See the milestones →

Case studies on this site cover personal production systems and home-lab infrastructure only — none of them describe an employer's environment, and no organizational details, identifiers or internal data appear anywhere on this site. The home lab is not a company environment — it is where deployment, identity, networking, observability and operational decisions get verified continuously.