← Back to Resume

Hyeonjun Kil

Backend Engineer / Platform Part Leader

kboxstar@gmail.com Seoul, South Korea

github.com/kilhyeonjun www.linkedin.com/in/kilhyeonjun Blog: kilhyeonjun.github.io

PROFESSIONAL SUMMARY

Backend engineer designing and operating a shared backend platform across 6 live game environments (12 branches), and leading the newly formed Platform Part to build team-wide AI development infrastructure. At GameDuo, drove an 82% BigQuery cost reduction, a 270x throughput expansion via serverless migration, and a WAL-sharded real-time game chat server (6,000 CCU target); presented the case study at Games on AWS 2025. Over 4 years and 8 months across four domains (Education, Healthcare, Clinical Trials, Gaming), consistently delivered DB optimization, concurrency control, and event-driven architecture design. Since late 2024, I have applied this same structural thinking to the development process itself — orchestrating multiple AI agents by role to design and ship production software. Right after the new Part launched, I designed and shipped a team-wide shared AI development context infrastructure (git-repo-based) that converts individual learnings into reusable team assets.

CORE COMPETENCIES

DB Optimization, Concurrency & Distributed Systems

Serverless Architecture & Cost Optimization

Legacy Modernization & Reliability

AI-Augmented Development Productivity

EXPERIENCE

GameDuo — DEV Team Platform Part Leader

2025.01 - Present

Mobile game development and publishing company. Shared backend platform design for 6 live game environments (12 branches), data pipeline engineering, and common package system operations

DEV Team Platform Part Leader (2026.04 - Present)

  • Newly formed platform team lacked task tracking and sizing framework. Designed 7-stage workflow (BACKLOG→SIZE REVIEW→DONE), handling rules for 4 task types, transition validators, and ticket template library. Standard operating framework deployed immediately upon team formation — task-type handling rules and decision criteria established simultaneously.
  • Existing sizing criteria did not reflect AI-assisted development time reduction with Claude Code. Established 5-dimension scoring framework and per-domain Claude Code time-discount rules (test writing ~80%, pattern replication ~70%). Team-consensus AI time-discount rules established, ensuring equitable performance measurement before and after Claude Code adoption.
  • Significant variance in AI agent productivity across team — context and learnings discovered by individuals did not propagate, causing repeated trial-and-error. Chose Git-based config-as-code over individual documentation — docs must be re-read by humans, but codified config integrates directly into the AI runtime and applies automatically. Built a team-wide AI development context platform unifying principles, coding standards, domain knowledge, agents, and hooks in one git repo. Established team-wide standardized AI context via git pull — converted individual learnings into team assets and eliminated AI productivity gap (397 domain knowledge files accumulated within ~2 weeks).

DEV Team Server Developer (2025.01 - 2026.03)

  • Marketing data query cost surge (full-scan billing model). Evaluated partitioning but rejected due to Write IOPS increase; adopted gRPC distributed reads (Storage Read API). Cut costs 82% ($6.25→$1.1 per TB).
  • Batch processing collection limits (60 days). Migrated from batch processing to EDA-based on-demand pipeline — distributing peak load and enabling throughput scaling. Expanded metrics range 6x (60→360 days), processing time 2h→5min.
  • Code duplication and convention inconsistency across 7 projects. Replaced manual code sync with shared NestJS package system (10 utility modules + Game Server Kit, auto-deploy). Automated upgrades cutting deployment time 3hrs→15min.
  • Marketing query latency (18s) from fragmented ad platform data (Google Ads/Meta/TikTok). Normalized read path + index tuning to avoid full-scan on 100-column table. Performance improved 97% (18s→0.5s).
  • Regulatory risk from game probability disclosure requirements. Integrated DynamicModule-based probability package across multi-game projects + CDK-based audit log pipeline. Achieved 94%+ integration test coverage (12 suites, 115 tests).

PROJECTS: Real-Time Game Chat Server (Multi-Tenant), Marketing Integrated Platform, Internal Common Library System

JNPMedi — Dev2 Team Docs Squad Backend Engineer

2023.06 - 2024.08

Clinical trial data management solution startup with KRW 16B cumulative investment (MSA-based)

PROJECTS: Maven Docs System Reliability Improvement, Maven VDR MVP Development

Medgo — Dev Team Backend Engineer

2022.04 - 2023.06

Telemedicine and medication delivery platform startup with 300K cumulative downloads

PROJECTS: Doctor/Pharmacist Web Service Enhancement, Backend Architecture Migration

SimpleHAN — Dev Team Backend Engineer

2021.07 - 2022.03

Custom software consulting firm

PROJECTS: Learning Management System (LMS) Development

Personal Projects (Side) — Platform / Tooling Engineering

2026.04 - Present

Self-directed platform/tooling engineering beyond day-job scope — an AI coding workflow automation framework and side projects

PROJECTS: Personal AI Coding Harness Framework

TECHNICAL SKILLS

Backend: Node.js, TypeScript, JavaScript, NestJS, Express.js, TypeORM, Java, Spring Boot, Jest, TDD, REST API, gRPC, Testcontainers, Integration Test, OAuth2 — Legacy-to-NestJS progressive migration, shared package system establishment, TDD-driven development, REST API / gRPC design

Database: MySQL, BigQuery, ElastiCache (Redis), Redis, Redis Streams, Valkey, ClickHouse — Query optimization, deadlock root cause analysis, large-scale table normalization, Redis distributed locking

Cloud & Infra: AWS (Lambda, ECS/Fargate, S3, SQS/SNS, CDK, EventBridge, Kinesis, Athena), GCP (BigQuery, Pub/Sub), Docker, GitHub Actions, CI/CD, Terraform, ARM64 (Graviton) — Batch-to-serverless migration, EventBridge/Kinesis event pipeline design, BigQuery cost optimization, CDK-based infrastructure as code, CI/CD pipeline

Development Productivity: Claude Code, Amazon Q Developer, Cursor, Context Engineering, Multi-Agent Orchestration, AGENTS.md — AI agent-augmented development, AGENTS.md-based context engineering, multi-agent orchestration, automated code quality

EDUCATION

Inha Technical College

Bachelor's (Advanced Major Program, Concurrent Employment) , Computer Information Engineering

2022.02 - 2023.02 | GPA: 4.10/4.5

Inha Technical College

Associate (Concurrent Employment) , Computer Information

2017.02 - 2022.02 | GPA: 4.16/4.5

CERTIFICATIONS

SQLD (Structured Query Language Developer) — Korea Data Agency, 2025.12

Engineer Information Processing — Human Resources Development Service of Korea, 2023.06

PRESENTATIONS & WRITING

Boosting Developer Productivity with Amazon Q Developer

Conference Talk — 2025.10

Games on AWS 2025 Customer Session

Email Delivery Monitoring through AWS SES Event Logs

Internal Seminar → Tech Blog — 2024.03

Presented internally then published as tech blog

TRAINING PROGRAMS

Hanghae Plus Backend 6th

2024.09 ~ 2024.11

Hanghae99

TDD and Clean Architecture, Concurrency Control, Redis

Week 1 Hall of Fame, Black Badge Completion

OPEN SOURCE

oh-my-openagent: Fix cached model context limits for Anthropic providers

2026.03

opencode: Fix macOS sidecar process hang on desktop app

2026.02

CodexBar: Support kiro-cli 1.24+ Q Developer format

2026.02

PROFESSIONAL DEVELOPMENT

System Design Interview Study

2025.08 ~ 2026.04

System Design Interview: An Insider's Guide Vol. 1 & 2

Node.js Design Patterns Bible Study

2025.10 ~ 2025.12

Node core concepts, concurrency, event loop, module system

Download