Lee ChanyangBackend Engineer

Language

← Back to design approaches

Process Evidence · 01

Classify requirements by purpose, actions, and constraints

Before translating feature names directly into APIs, I separated why each feature was needed, who would execute it, state changes, and policy, exception, and operational conditions.

I first broke down requirements from service stakeholders and domain documents by purpose, actors, state changes, and policy and operational constraints, rather than treating them as a feature list.

How I worked

Actual Working Approach

  1. Input

    I gathered requirements from service stakeholders, domain experts, planning documents, operational policies, screen flows, and the existing code and database structure in one place.

  2. Questions to ask

    I first asked what business purpose each requirement served, who would execute, approve, or view it, what state would change, and which policies and failure costs applied.

  3. Output

    I separated actor responsibilities, state changes, policy constraints, and exceptions into analysis units that could later inform domain models, use cases, and API contracts.

Source evidence

Cross-checking requirements-analysis evidence against actor-specific plans, use-case records, and production-service documents

I reviewed original work records alongside local archives and retained only evidence showing why something was needed, who changed state, and which constraints applied, rather than simply what was built.

B2B sample
Actor-specific plans, high-level specification reviews, and approval, settlement, and category requirements from a private B2B platform serve as samples of requirements classification.
Recurring practice
Earlier domain and use-case records show a practice of first writing relationship diagrams, sequence diagrams, success and exception scenarios, test cases, and input/output interfaces.
Production cross-check
Payment, subscription, refund, and cart documents show how screen requirements were separated into previews, orders, payment initiation, callbacks, status queries, and server-side revalidation.
Global requirements
In the representative educational-content service, I treated language, country, currency, and legal standards as separate requirement constraints so catalog display, terms acceptance, payment currency, and notification text would reference consistent criteria.
Exclusion criteria
Routine/Study Room records are used only as sources of judgment or learning records, not overstated as direct evidence for the requirements-analysis card.
Verification criteria

Preserving traceability from source requirements to models and use cases

Documents for one-time payments, initial subscription payments, refunds, and video access show that I documented purpose, the server's source of truth, state changes, exceptions, and frontend display criteria together, rather than simply converting screen flows into API lists.

Original requirements
I started with requirements visible in the UI, such as making a payment, starting a subscription, checking refund eligibility, and checking course access.
Classification results
I separated purchase intent, payment-gateway handoff, server-side revalidation, access grants, and refund execution into distinct responsibilities and state changes.
Portfolio presentation
I reconstructed the path from source requirements through analysis units to domain models and use cases as a decision flow suitable for public disclosure.
Detailed work example

Classifying B2B platform requirements by actor and operational responsibility

In the private B2B platform analysis, I compared actor-specific plans and high-level specification work with backend commit history, separating purpose, authority, state changes, and operational responsibilities before turning features into a backlog.

Classifying requirements by actor and responsibility boundaryMermaid

Rendering the diagram.

Where This Led to Implementation

  • Multi-Gateway Subscription Billing System · Payment and subscription requirement case map

    In this project, I placed video access, one-time payments, initial subscription payments, recurring renewals, and plan changes and cancellations in the same requirements queue, analyzing them as cases with distinct purposes, state changes, server-side sources of truth, and recovery points.

    View payment and subscription requirement cases
  • Points Accounting & Rewards System

    I kept the refund-eligibility check as a UX decision and separated actual refund execution into a state-changing use case that rechecks the original transaction, ownership, policy, and usage history and recalculates what must be restored.

    View the points usage and refund flow
Input
I read plans for platform operators, company administrators, employees, and partner vendors or small businesses alongside MVP priorities, MVP page structure, category proposals, and settlement and tax-invoice requirements.
Classification
I distinguished operators responsible for review, settlement, and visibility control; company administrators responsible for employee affiliation and benefit-fund configuration and reporting; employees responsible for verification, booking, and usage; and vendors responsible for onboarding, categories, bookings, and settlement history.
State changes
I separated registration approval, employee verification requests, vendor verification, category approval, pending and completed settlements, and tax-invoice issuance into flows requiring processing states such as PENDING, APPROVED, and REJECTED.
Output
From this classification, I derived Company, Vendor, UserAffiliation, UserRole, CertificationRequest, TaxCategory, and UICategory models, along with verification, approval, and rejection use cases.
Traceability evidence
The Service Build/Routine relationships verified through backend handover documents and original work records retain traces of analysis extending from plan review through domain modeling, overall relationship diagrams, use cases, sequence diagrams, and business flows.
Commit evidence
Git history shows changes connecting the domain, role, and affiliation model direction and employee-verification flow to controllers, DTOs, services, entities, factories, repositories, and Swagger documentation.
Limitations
This sample supports requirements analysis and modeling. It does not serve as evidence of a completed production service with successful runtime approvals, concurrency safety, or operational recovery.