Do Yeong Kim

I’m an AI Platforms Engineer at NAB. I helped deliver some of its first production AI agents, and I train and coach other engineers, because change happens from the ground up.

Melbourne, Australia

About

I work agent first: I direct, verify and own the outcome. The agents are ready. What isn’t ready is everything around them: the processes, the pipelines, the change processes, and the engineers who haven’t seen what an agent can actually do. Uplift starts with individuals and compounds into culture.

Experience

  1. AI Platforms Engineer

    Apr 2026 – Now

    NAB

    • Fast-release platform contact for four product teams taking AI agents to production in a one-month release; three have gone to prod. Dozens of PRs across 20+ repos and around 200 inbound questions.
    • On one team’s go-live day, cleared four production blockers in an afternoon and drove it to prod the same day.
    • Shared LLM gateway, intermittent 503s: found two causes. In test, one team’s load testing was giving everyone else 503s; they stopped and it went down 95%. In production, one manual batch job made over 99.9% of the calls. Built the gateway health dashboard as code; it showed a capacity increase had gone to the wrong model deployment, and another team building something similar decided to use mine instead of theirs.
    • An agent’s production restore job timed out and the first theory was credentials. Reasoned that a timeout means the network, found it 11 minutes into triage, and the fix was live in prod in two days.
    • Built my own agent workspace: 100+ repos, read-only cloud access, skills with safety evals, and hooks enforcing team coding standards. Ten engineers out of my team of about 30 adopted my starter kit in the first week.

  2. Associate Engineer

    Dec 2023 – Mar 2026

    Tech Academy, NAB

    • Ran about seven AI training sessions for around 300 people in total, including principal and senior engineers, and for the grad program. NPS 67; 11x velocity on production tickets over 3 days.
    • Codility booking app: automated a manual booking process for 8,000+ users with Power Apps and Power Automate.
    • Capability Matrix: page loads from 10–15 s to under 1 s across 200,000+ rows.
    • Confluence migration: 900+ pages with permissions kept; 30+ hours of manual work saved.

  3. Intern DevOps Engineer

    Jul 2022 – Dec 2023

    GOC Service Management & Automation, NAB

    • GOC Disaster Map, as product owner: crisis alerts from emergency-services feeds, supporting $5.2M+ in relief in the 2022–2023 floods.
    • AWS EC2 with Jenkins CI/CD; Airflow ingestion and scheduling; ITIL incident and change.

Projects

Fable Dash

A party game that makes my girlfriend really happy. She’s giving me constant feedback.

party.doyeong.kim Party game · web multiplayer

Duck, my personal SuperAgent, built in Amp

My superagent built on Amp. I press the action button on my iPhone, talk to Duck in English or Korean, and it does the work on my own machines. 158 tickets in about 3 days.

doyeong.kim/posts/how-i-built-duck Blog post

Education

  • Bachelor of Science, Computing and Software Systems 2024 University of Melbourne
  • AWS Certified Cloud Practitioner 2024 Amazon Web Services
  • International Baccalaureate: IB 44, Dux 2018 Geelong Grammar School

Skills

AI
Claude Code, Cursor, Amp, MCP, Amazon Bedrock / AgentCore, Databricks
Languages
Python, TypeScript, JavaScript, SQL
Cloud & data
AWS, Airflow, Elasticsearch, Docker, Jenkins, Harness
Microsoft
Power Apps, Power Automate, Power BI