👋 Welcome! A tech blog where a backend developer shares experiences in Backend, Infrastructure, Troubleshooting, and more ✨

About Me

Profile

“I’m Jeong Jeong-il, a developer who enjoys digging into the root of a problem, starting with the question ‘Why?’”

Hello! I’m a backend developer who loves uncovering the real story behind the error logs. For me, development is more than just writing code; it’s like a detective novel where I hypothesize and persistently dig to solve a single problem.

Instead of getting lost in the technical implementation itself, I enjoy asking myself questions like:

“What improvement does our service need most right now?”
“What’s the optimal way to solve this problem, and is it right for our team’s situation?”
“What do our users really want?”.

I believe these considerations act as the most solid compass to guide my technical decisions.

This blog is a record of my struggles and growth. Rather than glamorous success stories, I want to honestly share the deep thought/worry I faced during problem-solving and the ‘Aha!’ moments of realization. Please feel free to share your experiences and opinions. I want to share the joy of learning and growing together.

GitHub 잔디

Skills

SkillExperience (Years)Description
Java3.0Experience developing backend systems in various environments such as MSA and Batch
Kotlin2.5Experience with asynchronous processing using Coroutines and as a primary language in MSA environments
Spring Boot3.0Experience developing and operating REST APIs using Spring Security, Actuator, etc.
Spring Batch3.0Experience designing large-scale data processing systems and optimizing performance through Chunk and Partitioning
Spring Cloud2.0Experience building and operating MSA using Gateway, Eureka, etc.
MySQL3.0Certified - SQL Developer | Experience in database design, query tuning, and Replication operations
Kubernetes2.0Certified - Certified Kubernetes Administrator (CKA) | GKE infra design, Helm package management, and ArgoCD deployment automation
Linux4.0Certified - Linux Master Level-II, AWS Certified Cloud Practitioner (CCP) | Experience in server configuration, script writing, and operations

Work Experience

Bootalk Inc.

AI-based and BigData-utilized real estate brokerage platform
2025.03 ~ Present ( 9 months ) | Dev. Team - Back-end Engineer

[Key Technologies Used]: Kotlin Spring Cloud Kubernetes ArgoCD Opentelemetry Grafana LGTM Stack Redis MySQL Manticore Search

  • Participated in Design and Implementation of Monolithic Architecture → MSA Transition

    [Key Achievement] Doubled the speed of new feature additions by decoupling tightly coupled domains and successfully transitioning to MSA
See Details
  • Project Overview: A project that migrated a Java Servlet legacy monolithic architecture service to a Kotlin/Spring Cloud-based MSA with zero downtime
  • Duration: 2025.04 ~ 2025.09 (3 members)
  • [Problem]
    • At the time of joining, MSA separation of loosely coupled domains such as notifications and user events was complete.
    • Tightly coupled core domains (real estate listings, broker-user matching) encountered strong coupling problems that could not be solved by existing strategies → Gradual transition was impossible.
    • Existing MSA transition strategy (separate separation by domain) carried risks of data inconsistency and tangled business logic.
  • [Solution & Role]
    • Established and executed a realistic MSA transition strategy to resolve tightly coupled domain issues, and designed transition period management strategies.
      • Applied a partial big bang approach by grouping tightly coupled domains (real estate listings + user matching) as a single transition unit for simultaneous separation.
      • During the transition, established a new MSA API call structure from legacy, dual writing for data duplication, and batch processing.
    • Removed external technical dependencies from domain logic with Hexagonal Architecture + CQRS pattern.
    • Reviewed and discussed various communication methods such as gRPC and Zero-Payload, built synchronous communication between services based on FeignClient, and event processing based on AWS SQS.
  • [Achievements]
    • Achieved service flexibility and independent scalability by operating independent service units per domain through MSA transition.
    • Successfully migrated to MSA while maintaining 100% service availability of core services through zero-downtime transition.
    • Increased the speed of subsequent service expansion and new feature additions by approximately 2 times by stably separating tightly coupled domains.
  • [Key Technologies]: Kotlin, Spring Cloud, gRPC, Feign Client, CQRS, Hexagonal Architecture, AWS SQS, AWS SNS, Redis

  • Built and Operated Kubernetes-based Infrastructure and Monitoring System

    [Key Achievement] Led Kubernetes transition to secure high availability and scalability, and built a monitoring system to reduce incident detection time by 98%.
See Details
  • Project Overview: A project that proposed, designed, and built a Kubernetes-based transition from a single VM environment and proposed, designed, and built a monitoring system to secure high availability and operational efficiency.
  • Duration: 2025.04 ~ 2025.09 (1 member)
  • [Problem]
    • Single instance + Shell Script/Docker CLI based deployment environment → Single Point of Failure (SPOF) leading to decreased service reliability, requiring developer intervention for service expansion and disaster recovery.
    • Absence of monitoring system led to reliance solely on user CS for incident detection → Average of 1 hour to recognize problems.
  • [Solution & Role]
    • Proposed, designed, and built Kubernetes-based orchestration to secure high availability and service scalability.
      • Compared Managed Kubernetes (GKE) vs Self-Managed Kubernetes, then selected and introduced GKE considering in-house resources.
      • Removed Spring Cloud Eureka, API Gateway dependencies → Transitioned to Kubernetes native environment.
      • Built deployment automation with GitHub Actions based CI/CD pipeline and ArgoCD.
    • Recognized the 1-hour incident resolution time as a critical issue that could lead to service abandonment, and proposed, designed, and built a monitoring and incident detection system.
      • Compared ELK Stack vs Grafana·Loki·Tempo·Prometheus technologies and reviewed cost efficiency, then proposed and built the latter.
      • Centralized log/metric collection using Kubernetes sidecar pattern.
  • [Achievements]
    • Incident Detection Time: Average 1 hour → 1 minute (98% reduction), proactive incident detection and response possible without user CS, improved operational efficiency and productivity through metric visualization.
    • Eliminated developer intervention in incident recovery and service expansion → Secured automation system.
    • Rollback Time: Average 12 minutes → 2 minutes (83% reduction).
    • Secured zero-downtime deployment system with Kubernetes rolling updates, improved deployment stability.
  • [Key Technologies]: Kubernetes, GCP GKE, ArgoCD, Github Actions, Grafana, Loki, Tempo, Prometheus, Opentelemetry, Kubernetes Sidecar

  • On-premise → AWS Cloud Zero-Downtime Transition

    [Key Achievement] Successfully migrated from on-premise to AWS cloud with zero downtime without service interruption.
See Details
  • Project Overview: A project that migrated services operating in an on-premise IDC to an AWS cloud environment with zero downtime to improve stability, scalability, and operational efficiency.
  • Duration: 2025.07 ~ 2025.08 (2 members)
  • [Problem]
    • Intermittent server downtime issues due to hardware contact failures in the on-premise environment.
    • Limitations of manual scaling due to increasing traffic and increased operational burden due to complex operating environment.
  • [Solution & Role]
    • Responsible for establishing and executing a migration strategy for zero-downtime transition: Established a dual operation strategy considering DNS propagation time to minimize degradation of user experience.
    • Resolved data consistency issues (introduced CDC)
      • Reviewed various technologies (Replication, Dual Write, Message Queue, etc.) to resolve data inconsistency issues between IDC DB and AWS RDS during the migration period.
      • Introduced CDC (Change Data Capture) technology, which has high real-time performance and reliability, to build a unidirectional real-time synchronization pipeline from IDC DB to AWS RDS. For the CDC solution, AWS DMS (Data Migration Service) was chosen for its excellent compatibility with RDS and stable managed service advantages for migration to the AWS environment.
    • Infrastructure automation (Terraform IaC): Automated core infrastructure resources such as EC2, RDS, VPC in the AWS environment with Terraform code.
  • [Achievements]
    • Successful zero-downtime transition: Successfully migrated from on-premise to cloud without service downtime, maintaining user experience.
    • Improved operational stability: Resolved server downtime issues due to hardware failures.
    • Secured scalability and efficiency: Achieved scalability to flexibly respond to traffic changes through transition to the cloud environment.
    • Minimized human error through IaC, securing infrastructure reproducibility and documentation.
  • [Key Technologies]: AWS, Terraform, AWS DMS(CDC), AWS RDS, AWS EC2, AWS VPC

Hello Fintech Co., Ltd.

P2P online investment-linked financial platform startup that achieved 2.5137 trillion won in cumulative loans
2023.03 ~ 2025.03 (2 years) | Backend Developer

[Key Technologies Used]: Java Spring Boot Spring Batch JPA QueryDSL Redis MySQL Kubernetes ELK Stack

  • Settlement Renewal Project

    [Key Achievement] Integrated settlement logic distributed across 3 servers, 24 DB Procedures, and DB Functions, improving requirement reflection speed by 46%.
See Details
  • Project Overview: Proposed, designed, and built integration of settlement business logic that was previously separated into multiple domains.
  • Duration: 2024.06 ~ 2025.03 (4 members)
  • [Problem]
    • Settlement logic was distributed across 3 servers, 13 DB Procedures, and 11 DB Functions, leading to high complexity and excessive maintenance and development costs.
    • Race Condition occurred in distributed environment.
  • [Solution & Role]
  • [Achievements]
    • Reduced system complexity leading to an average 46.31% improvement in requirement reflection speed (4.06 weeks → 2.18 weeks).
    • Improved reliability and stability of settlement process by resolving concurrency issues.
    • Successfully integrated while maintaining service operational stability through gradual transition.
  • [Key Technologies]: Spring Boot 2.7.3, Java 11, JPA, Redisson, Rest API

  • Developed, Operated, and Improved Back-office Web Server

    [Key Achievement] Improved major API performance by 87% through SQL tuning and caching.
See Details
  • Project Overview: Renewed PHP legacy back-office server to Java, Spring, and further improved existing technical debt.
  • Duration: 2024.04 ~ 2025.03 (2~6 members)
  • [Problem]
    • HikariCP Deadlock occurred due to a mix of JPA and MyBatis within one Task.
    • Data inconsistency issues due to DB Replication lag.
    • Increased server load due to inefficient subqueries, unutilized indexes, and N+1 problems.
    • Performance degradation due to mixed domain logic in a single logic (e.g., saving evaluation results, sending SMS, administrator notifications when approving a loan).
  • [Solution & Role]
    • Explored and identified the cause of HikariCP Deadlock, resolved through unification to JPA.
      • Considered separating connection pools, changing connection return timing of persistence context, but since Mybatis code was already scheduled to be changed to JPA, resolved through unification to JPA.
      • Prioritized fixing the problematic logic, and during the change week, increased the number of Pods to distribute load to prevent DBCP connection exhaustion.
    • Resolved DB Replication lag issue
      • By dynamically separating DataSources using AbstractRoutingDataSource and AOP, the system was configured to use Slave DB for data retrieval and Master DB for changes, resolving data inconsistency issues caused by Replication lag.
    • Performance Improvement
  • [Achievements]
    • Improved user reliability by resolving replication lag.
    • Improved server stability by resolving HikariCP Deadlock issues.
    • Improved major API performance by 87.66% through SQL tuning (Average: 10.295 seconds → 1.27 seconds).
    • Improved API response speed by 74.69% through Redis-based caching (Average: 5.132 seconds → 1.299 seconds).
    • Improved system maintainability and performance by separating concerns through event processing.
  • [Key Technologies]: Spring Boot 2.6.7, Spring Security, Java 11, JPA, Mybatis, Thymeleaf

  • Developed, Operated, and Improved Batch Server

    [Key Achievement] Reduced batch data processing time by 64% by introducing Chunk/Partitioning.
See Details
  • Project Overview: Renewed PHP, Crontab legacy Batch server to Spring Batch, and further improved technical aspects and operations.
  • Duration: 2024.05 ~ 2025.03 (4~6 members)
  • [Problem]
    • Built with PHP and Crontab, making management difficult, and Deadlock issues occurred when Job ran concurrently during the transition to Spring Batch.
    • Performance degradation and data consistency issues when processing large amounts of data with Tasklet method.
  • [Solution & Role]
    • Identified and resolved Spring Batch Metadata Table Deadlock cause
      • Considered methods such as prohibiting concurrent execution of multiple Jobs, upgrading Spring Batch version (rejected due to requiring full Java, Spring Boot, Spring Batch version upgrade), and changing Isolation Level related to Metadata Table.
      • Prohibiting concurrent Job execution was unrealistic and rejected, Spring Batch version upgrade was rejected due to requiring full Java, Spring Boot, Spring Batch version upgrade, so resolved by changing Isolation Level related to Metadata Table.
    • Introduced Chunk-based processing and Partitioning method to improve large data processing performance.
      • Optimized large data processing performance by compensating for the shortcomings of the Tasklet method.
      • Significantly improved batch performance from 13 minutes to 5 minutes.
  • [Achievements]
    • Established a stable Batch environment by resolving Deadlock issues.
    • Reduced data consistency issue occurrence by 99.9% by resolving data consistency issues (maximum transaction processing time 22 minutes → 0.01 seconds).
    • Optimized performance by reducing data processing time by 64% (13.27 minutes → 4.77 minutes).
    • Shared resolution case within the company, documented for use by other teams.
  • [Key Technologies]: Spring Boot 2.7.3, Spring Batch, Java 11, JPA, MyBatis

Activity

Certificate