Company

Problem Recommendation System Upgrade

Reverse-engineered an undocumented legacy PHP recommendation system and rebuilt its data structure and logic

Period | 2025.03 ~ 2025.06

  • #Python
  • #MySQL
  • #PHP
  • #Git
  • #Notion

Stabilizing and restructuring an undocumented legacy system

Problem

  • No system specification or data dictionary
  • Data duplicated across 92 tables
  • Presentation, business and DB logic tightly coupled within PHP files
  • Non-standard rules, duplicated logic and hard-coded parameters made it impossible to change or experiment with the recommendation logic

My Role

  • Analyzed core FastAPI endpoints for recommendation, grading and roadmap generation, and mapped the full logic flow
  • Merged level-calculation and recommendation algorithms implemented in different ways into one module
  • Traced schemas and I/O data across 92 tables to rebuild the log and DB mapping
  • Tracked and fixed defects such as SQL injection vulnerabilities and column name mismatches in Notion

Result

  • Documented structure and data flow, improving maintainability
  • Established an environment where recommendation logic can be changed and tested
Recommendation process and data flow analysis
Recommendation process and data flow analysis
Issue tracking and fix history
Issue tracking and fix history