Skip to main content Link Menu Expand (external link) Document Search Copy Copied
MinStack

Hello, I’m Gyumin Hwangbo, a backend engineer who has handled up to 4,000 RPS across Kyochon’s order, payment, and membership services and has designed, operated, and automated highly available systems with Spring Boot and Kotlin.

More recently, I have been using Codex for code analysis, system design, implementation, refactoring, testing, and documentation, expanding an AI agent-based workflow. I have also built CDC pipelines, improved large-scale batch performance by 95.4%, created per-brand CI/CD pipelines, and designed Redis Streams-based messaging systems to improve operational automation and system efficiency.

I have investigated the causes of performance bottlenecks and missed notifications in production and improved systems to prevent those problems from recurring.

EXPERIENCE

  • Backend Engineer (FOODDASH) 2024.04.15 ~ 2025.11.30

    • Developed and operated order, payment, and membership backends serving 6M+ members across multiple franchise brands, including Kyochon with 770k MAU, while handling peak Kyochon traffic of up to 4,000 RPS and supporting zero-downtime deployments.
    • Migrated Jadam service data from Wmpo and stabilized operational issues after the service transition.
    • Analyzed bottlenecks in the membership grade renewal batch for 1M+ users and reduced processing time from 19m13s to 53s (95.4% improvement).
    • Separated brand-specific features from shared modules and built an integrated CI/CD pipeline with change-based staging deployments, manual production triggers, rollout tracking, and Slack notifications.
    • Investigated notification loss caused by external Push API timeouts, retries, and server restarts when large notification batches overlapped with real-time order notifications.
      • Served as the technical lead for introducing a Redis Streams message queue to solve notification loss, using Consumer Groups and PEL (Pending Entries List) to provide at-least-once delivery semantics, track in-flight messages, and recover unprocessed notifications after restarts.
      • Stabilized notification loss to near 0% (link).
    • Designed and operated a MySQL CDC -> MQ -> ETL -> MongoDB synchronization pipeline to support fast, large-scale real-time reads by denormalizing read models and reducing multi-query bottlenecks across order and menu tables.
    • Stack: Kotlin, Spring Boot, MySQL, Redis Streams, Kubernetes, AWS

  • Senior Researcher (foxee) 2023.06 ~ 2024.01

    • Built backend APIs for a vulnerability analysis web service with Java and Spring Boot.
    • Conducted CNN/XAI research for malware image analysis and built preprocessing/visualization pipelines for the Windows 2015 malware dataset.
    • Improved the development and deployment workflow with Docker and GitHub Actions.
    • Stack: Java, Spring Boot, PostgreSQL, Docker, Python

  • Intern (Pulse Co., Ltd.) 2020.01 ~ 2020.02 (1 month)

    • Developed internal workflow support tools with Python.

EDUCATION

OPEN SOURCE

ACTIVITIES

  • Patent: Blockchain client vulnerability detection method and device (2022.01 ~ 2022.12)
  • Conference: (NLP) N-gram + threat-actor-based pattern analysis for Windows malware families (2021.12)
  • Software registration: Blockchain-based REC trading platform prototype for RE100 [C-2021-044149] (2021.09 ~ 2021.11)
  • Hackathon: Convergence Security Graduate School Hackathon Competition (2021.09 ~ 2021.11)
  • AI Competition (4th): 2021 Cybersecurity AI/big data challenge, KISA — fileless malware detection (2021.09 ~ 2021.12)
  • Privacy Competition (3rd): 2020 K-cyber security challenge, KISA — privacy (2020.09 ~ 2020.12)
  • Overseas training: San Jose State Univ., San Jose, CA, USA (2019.06 ~ 2019.09)
  • Exchange student: HELP Univ., Kuala Lumpur, Malaysia (2017.12 ~ 2018.03)

CONTACT