We’ll Fill the Gap Instead of Experience — Practical System Design for Juniors in the AI Era
Become a backend developer who can explain “why it made that decision” about AI-generated code. Across commerce, payments, advertising, and deployment, we’ll turn the design decisions seniors actually made in the field into 8 modules, real fines and outage cases, and hands-on practice with starter code.
Designing Redis Atomic Decrements to Achieve Zero Overselling Under 100,000 Simultaneous Requests for First-Come, First-Served Coupons
How to Design a Payment API That Charges Only Once Even When Retries Surge (Idempotency Keys · Header Versioning · 6 Status Code Types)
The Principle Behind an Ad Settlement Microbatch That Doesn’t Miss Even One Won at 100,000 Events per Second
Completed a multi-module mini-shop project where the compiler enforces module boundaries (internal·DIP)
Executing blue-green deployments and rollbacks on local Kubernetes and building Prometheus·Grafana dashboards
Architecture decision criteria for choosing “only as much as you need right now” at the initial, 100K, 1M, and 10M stages
A review habit of constraining and validating AI-generated code based on compliance, cost, and scale criteria
Complete a Request Logger with Automatic PII Masking + traceId Tracking (Starter Code Exercise)
AI가 짜준 코드, 면접에서 설명하실 수 있나요?
This is an era in which AI writes code for us. Features are built much faster than before. Yet in interviews and code reviews, people ask questions like this.
"Why did you decide to write the code this way?"
The opportunity to build up the practical experience that once would have taken years of trial and error has itself diminished. AI gives you a plausible answer when you ask, but it does not make three judgments for you.
Whether this code violates the Personal Information Protection Act
How much this architecture incurs in cloud costs
Whether this is overengineering for our current scale
This course fills that gap. It presents the design decisions I actually encountered in commerce, fintech, and mobility, along with what I organized while mentoring Software Maestro trainees. You will learn not “what to use,” but “why seniors choose it.”
이런 분께 추천해요
① Backend developers with 0–3 years of professional experience
Those who can build features but cannot explain "why they designed them this way." Those who get stuck when asked, "Why did you do it this way?" in code reviews.
② Those preparing for employment or a career change
Those who want to explain the reasoning behind why a real company made a particular choice, rather than memorize answers for system design interviews.
③ For those who use AI coding tools every day
For those who feel uneasy about simply pasting code from Copilot, Claude, or ChatGPT while wondering, "Is this right?" and want to independently assess the compliance, cost, and scalability considerations that AI overlooks.
After completing the course, you’ll gain the following abilities.
Decision criteria for handling requests that 100,000 people make simultaneously—such as first-come, first-served coupons and inventory—using DB locks, Redis atomic operations, distributed locks, and Kafka queues step by step.
How to design payment APIs by compromising with reality rather than following standards to the letter, and how to prepare for the six points of failure when traffic surges tenfold
How to prevent fines and costs caused by a single line of logs with code (personal data masking, preventing cost explosions)
A roadmap for deciding what to adopt and what to defer at each service scale (initial → 100,000 → 1 million → 10 million)
Build a multimodule architecture that lets the compiler enforce boundaries, and see firsthand how that architecture integrates with AI coding agents.
The pitfalls of zero-downtime deployments (rolling, blue-green, and canary), and incident monitoring and alert design based on the Four Golden Signals
Features of the Lecture
1. We show you the decision-making process, not just the right answer
Every module follows the sequence: "Why does this problem occur → How was it assessed in the field → How much of it should we adopt for our scale?" Module 4 also includes a separate critical reflection lesson titled "Can we apply this structure directly to our project?".
2. Experience it through real-world cases
We cover real regulatory and outage patterns, such as personal information fines of 7.5 billion, 135 billion, and 3.5 billion won, replication delays under 10× traffic, and connection pool exhaustion during a first-come, first-served event. The cases have been anonymized and reconstructed without identifying specific companies.
3. Understand Through Diagrams
We explain using over 40 sequence diagrams, flowcharts, and architecture diagrams, pointing things out with “Look at the left side of the diagram.” No concept is covered only verbally and then left behind.
4. Build it yourself with starter code, then verify it yourself with AI
Modules 2 (logging), 7 (multi-module), and 8 (deployment and monitoring) are built hands-on using starter projects containing TODOs. Every module includes a hands-on exercise box called “Try giving the same prompt to AI and compare,” allowing you to identify gaps in AI’s answers for yourself.
What you’ll learn
Module
Topic
Key Question
1
Orientation
Why Is “Judgment Criteria” Experience in the Age of AI?
2
Logging Architecture in the AI Era
Why a Single Line of Log Code Written by AI Can Lead to Fines and a Cost Explosion
How to prevent overselling when 100,000 people rush to buy 100 items in stock
4
Payment API Design and Handling Traffic Surges
Why Did Major Payment Gateways Abandon PUT/DELETE, and What Breaks When Traffic Increases Tenfold?
5
Advertising Settlement Engine
How do you trade off performance and consistency at tens of thousands of TPS?
6
Architecture by Stage: From 0 to 10 Million
What is overengineering at our current scale?
7
Domain-Centric Multimodule (Hands-on)
To entrust the rules to the compiler rather than people
8
Zero-downtime deployment and incident monitoring
Why things break with every deployment and how to avoid waking up the team in the middle of the night
Why I Created This Course
In teams without a mentor, we live in an era where asking AI produces an answer. But I’ve seen far too many teams adopt plausible-looking architectures wholesale because no one was there to judge whether “this is right for our situation,” only to have it come back as costs and incidents. I wanted this course to provide the experience of having a senior beside you say, “That’s overkill for our current scale.”
This course does not cover these topics.
The basic syntax of Kotlin, Spring Boot, and SQL (assumed as prerequisite knowledge)
Coding tests and algorithms, frontend development
Infrastructure setup in general, such as creating an AWS account or setting up a cluster (deployment exercises are based on a local Kubernetes environment)
Actual internal code from a specific company — the examples in Modules 4 and 5 are fictional scenarios that synthesize patterns from multiple systems.
Notes Before Taking the Course
Practice Environment
IntelliJ IDEA, Kotlin + Spring Boot, Docker (required)
Local Kubernetes: Docker Desktop or minikube (Module 8 deployment practice)
Redis and Kafka can be run with Docker or followed through the code explanations. No prior experience required.
An AWS account is optional. If you use one, be sure to delete the resources after the hands-on exercises.
Slide PDFs for all modules, and hands-on progress notes (step-by-step guidance based on file paths and line numbers)
Modules 3–6 are “hands-on theory” lessons where we read the code together and follow the reasoning. The Hands-on modules, where you type the code yourself, are Modules 2, 7, and 8.
A total of 8 modules and 46 lessons, approximately 6 hours. In October, 7 AI Hands-on lessons (approximately 1 hour) will be added for free.
Prerequisites
Basic Kotlin or Java syntax (Knowing only Java is sufficient. The code is short and accompanied by explanations.)
Experience creating an API with Spring Boot at least once
SQL basics (what SELECT, INSERT, and transactions are)
Things to Know
The AI answers shown in the course reflect the interface as of September 2026. If the model changes, the answers will also change.
For penalty and outage cases, only publicly disclosed enforcement facts are retained, and company names have been anonymized.
Updates planned
We will continue adding to the course according to the schedule below even after launch. The additional lessons will be provided free of charge to existing students.
Late September 2026: Launch of an AI hands-on challenge
By mid-October 2026: Add 7 module-specific AI hands-on lessons — instead of typing the code yourself, you give an AI coding agent the module’s decision criteria as constraints and have it implement the solution, then verify it with k6, tests, and metrics and critique it according to senior-level standards (starter code, docker compose, and verification scripts provided)
Starting in November 2026: Module-specific Q&A lessons based on actual questions from the student question board
Instructor Introduction
In my sophomore year of college, I got a game I built on my own into the upper ranks of the App Store, then founded a company as a developer and CEO and ran it for five years. Later, as a freelancer, I led the development of PeaceMinerOne’s early e-commerce system and served as CTO at Newploy. In the meantime, “MERS Map,” which I built on my own, grew to 1 million DAU and was covered by media outlets in Korea and abroad.
I joined Toss before it became a unicorn and built its initial advertising system, loan brokerage, insurance and automotive-related systems, and payment API. At Toss Payments, I was a founding member and built the payment system from the ground up. I later led e-commerce development for brands such as Carhartt and Worksout, worked on high-traffic systems at Kakao Mobility, and served as a founding member and CTO at Voltup, a joint venture between LG Uplus and Kakao Mobility, where I led the initial design and development.
As a Software Maestro mentor, I taught trainees a course that served as the basis for this lecture.
Recommended for these people
Who is this course right for?
Backend developers with 1–3 years of experience who use AI to build features but get stuck answering, “Why did you do it this way?” in code reviews
A junior developer who was assigned the first-come, first-served and payment features but feels anxious because they’ve never actually experienced a concurrency issue.
Job seekers and career changers preparing for system design interviews who get stuck at “scale assumptions” (with Spring Boot project experience)
An early-stage developer at a startup with no senior engineers on the team, handling logging, deployment, and monitoring by gut feeling.
A solo developer with no one to ask whether their side project’s cloud costs and handling of personal information are okay.
Need to know before starting?
You need to know the basic syntax of Kotlin or Java. Knowing only Java is sufficient. The code is short and accompanied by explanations.
Experience creating an API with Spring Boot at least once is required.
Basic SQL (SELECT, INSERT, and what transactions are) is sufficient.
The hands-on exercises require IntelliJ IDEA and Docker Desktop. The Module 8 deployment exercise uses local Kubernetes (Docker Desktop’s Kubernetes or kind).
AI coding tools can be anything, such as Claude, ChatGPT, or Cursor, and you can follow along with the free versions. The AI Hands-on lessons added in October use AI coding agents.
An AWS account is optional. If you use one, be sure to delete the resources after the lab.
No prior experience with Redis, Kafka, or Kubernetes is required. We’ll either run them with Docker or follow along with code explanations.
🏦 Former early member of Toss · Founding member of Toss Payments | Backend developer with experience developing advertising, loan, insurance, and payment APIs
🚗 Former) Kakao Mobility large-scale traffic systems · Founding member and CTO of BoltUp, a joint venture between LG Uplus and Kakao Mobility
🛒 Former e-commerce development lead at Peace Minor One, Carhartt, and Worksout · Former CTO of Newploy
🚀 Former undergraduate startup CEO for 5 years (2014 Red Herring Asia Top 100 Startup) · Solo-developed “MERS Map,” reaching 1 million DAU and covered by domestic and international media · Indie game developer ranked in the top 10 in the App Store Games category (2010)
🎓 Software Maestro Mentor | Conducting 0-1 mentoring and system design classes
🤖 In the age of AI coding, I teach juniors not “what to use,” but “why seniors choose it.”
![[Backend/Exception Handling Scenario/Aggregation Optimization] Backend Portfolio and Practical Experience Enhancement Strategy. All-in-One PART1Course Thumbnail](https://cdn.inflearn.com/public/courses/335091/cover/1a19a4de-ec2e-4e26-a84e-28691e777020/335091.jpg?w=420)
![[Season 1] Spring AI in Practice: Developing a Premium Reservation AI AgentCourse Thumbnail](https://cdn.inflearn.com/public/files/courses/339732/cover/01kb2qr7zw5gf0724jvv4rndns?w=420)
![[4-Week Completion] Clean Architecture with AI for Senior Developer AdvancementCourse Thumbnail](https://cdn.inflearn.com/public/files/courses/339355/cover/01k8ags08xj5tpgyrpv77hkj21?w=420)
![[Season 2] Spring AI in Practice: Developing Multi-AI Agent SystemsCourse Thumbnail](https://cdn.inflearn.com/public/files/courses/339915/cover/01kbqgcec3hb576vxxdf9ch76y?w=420)
