Lee ChanyangBackend Engineer

Language

← Back to project list

Legacy Platform Renewal

Attale & Attale Pro Life Planning Platform Renewal

I analyzed both the consumer life planning service and its expert and operator counterpart, then redesigned the API, database, and content migration structure.

Company
Life Planning Lab
Service
Attale / Attale Pro
Period
2022.03–2022.05 · 3 months
Role
Backend Engineer
Responsibility
End-to-End Renewal · User/Pro API · DB · Migration · AWS
  • TypeScript
  • Express
  • TypeORM
  • MySQL
  • AWS
  • S3
  • Swagger
  • Docker

Project Overview

I redefined goal and content boundaries between Attale and Attale Pro

The core challenge was keeping Attale's user goal execution flow and Attale Pro's recommended-goal and content management flow from becoming mixed in the same data structures.

Within the backend renewal scope, I analyzed the legacy structures of both services and separated curated originals, user goal instances, the goal → task → daily hierarchy, eight life areas, and the hashtag and content graph. I migrated operational content from CSV/Notion into the new table structure and documented the Express/TypeORM APIs together with evidence of an operational transition based on S3, Swagger, and Docker.

  • I redesigned the shared data boundaries around Attale's user goal execution and scheduling flow and Attale Pro's recommended-goal and content management flow.
  • Curated goals remain as originals, while goals started by users become separate instances with individual progress states.
  • I separated starting a goal through copying from simply saving content through a link, distinguishing access to current originals from the ability to edit a personal copy.
  • I separated the publicly shareable evidence of cleaning and migrating source content comprising approximately 307 recommended goals, 1,052 tasks, and 81 articles.

Ownership

My direct responsibilities

  • Analyzed Attale's user goal execution flow alongside Attale Pro's recommended-goal and content management flow
  • Redesigned the Express/TypeORM API and database structure shared by both services around the new product requirements
  • Implemented separate flows for copying a recommended goal into a personal goal and linking to original content when saving it
  • Modeled the goal → task → daily hierarchy, eight life areas, hashtags, and content graph, and migrated the data
  • Established foundations for the operational transition, including authentication, S3 uploads, Swagger documentation, and Docker/ECR deployment scripts

Constraints

Constraints to address first

  • Attale for consumers and Attale Pro for experts and operators used the same goal and content data, but differed in who edited it and how it was used.
  • Operator-edited curated originals and personal goals that users pursued and edited could not share the same lifecycle.
  • Starting a goal and saving content for later had different storage costs and requirements for staying current.
  • The eight life areas and detailed categories needed to be managed as operational data rather than code enums.

Implementation Timeline

How I separated user execution from operational content management across the two services

I reviewed the legacy structures of Attale and Attale Pro together, then separated and migrated user execution data and operational content management data. Select a phase to see the problem identified at that stage and how it led to the architecture and verification criteria.

Phase 01 · Attale & Attale Pro Data Boundaries

Attale & Attale Pro Data Boundaries

Phase 01

I separated the data lifecycles of consumer-facing Attale and expert-facing Attale Pro

Although they may look like the same goal, a recommended goal managed in Attale Pro and a goal a user actively pursues in Attale differ in who edits them, how current they need to be, and when they are deleted.

Attale & Attale Pro Data BoundariesMermaid

Rendering the diagram.

  1. 01

    Requirements Analysis

    Attale Pro manages recommended goals, articles, and videos; Attale users bring these into personal goals and add their own schedules and progress states.

    • If user edits altered the original template, incorrect content could spread to other users.
    • User goals needed individual state such as duration, budget, completion status, and child tasks.
  2. 02

    Domain Modeling

    I separated curated goals, articles, and videos as original content from user goals and saved content as user data.

    • Operators manage original content, which multiple users reference.
    • User data has its own owner, progress state, deletion status, and subordinate plans.
  3. 03

    Architecture Decisions

    Instead of mixing original content and user data in one table with a status field, I separated them into copying and linking paths.

    • Editable goals require copies, while bookmarks intended to show the latest original need only a link.
    • This distinction became the basis of the subsequent template-to-instance model.
  4. 04

    Implementation & Verification

    The archived entities and ADRs show curated originals, user goals, and user-to-content link models.

    • I distinguished active entities from commented-out design files to recheck what was actually implemented.
    • Service names are used to explain the scope, but internal API paths and operational data were removed from public copy.
  5. 05

    Operations & Recovery

    When data becomes inconsistent, the first step is to distinguish whether the original content changed, the user's copy changed, or only a link remains.

    • Without that distinction, an operator's edit to an original could unintentionally affect user goals.
    • Deleting user data must have different recovery criteria from deleting original content.

Project Infrastructure

I established transition foundations spanning authentication, file uploads, documentation, and deployment scripts

I organized evidence of the runtime, storage, authentication, file uploads, documentation, and deployment supporting the API transition for both services into shared infrastructure boundaries.

  1. 01

    Runtime

    The stack includes TypeDI services, TypeORM repositories, and Swagger documentation on Node.js/TypeScript and Express 4.

  2. 02

    State Storage

    MySQL and TypeORM entities handle goals, content, classification, and user data.

  3. 03

    External Integrations

    The package stack retains email registration, JWT and bcrypt authentication, Kakao and Apple OAuth, an S3 upload library, Swagger, and logging/security middleware.

  4. 04

    Deployment Evidence

    Docker build and ECR push scripts remain, but actual deployment logs, server topology, and operational documentation could not be verified.

Outcome

Verifiable outcomes

  • Defined the API and database renewal scope around both Attale's user goal execution flow and Attale Pro's recommended-content management flow
  • Defined a data model separating template-to-user-goal copying from bookmark links
  • Migrated recommended-goal, task, and article source content from CSV/Notion into a new relational structure
  • Modeled the goal → task → daily hierarchy and category/content graph
  • Established Router → Controller → Service → Repository → Entity layers with Express, TypeORM, and TypeDI
  • Verified evidence of an operational transition through email/JWT/OAuth authentication, S3 image uploads, Swagger documentation, and Docker/ECR scripts
  • Reduced the risk of overstating an older project by distinguishing repository dates from snapshot verification criteria

Limitations

Limitations and unimplemented scope

  • Service names are used where publicly shareable, but private repository URLs, operational data, authentication files, and internal API paths are not disclosed.
  • Only a single merge snapshot can currently be verified in the local original repository; the detailed sequence of 33 commits could not be reverified.
  • Response-time measurements and improvements based on AWS bills from that period are listed on my resume. Because the original logs and billing records are no longer preserved, this detail page publishes only evidence of structural changes.
  • Commented-out vision, keyword, asset, and report-image designs are not labeled as completed implementations.
  • The deployment pipeline and server topology are not described in concrete configuration detail because publicly shareable source evidence is insufficient.