I designed the algorithm using SQL queries and Python for data processing. First, I normalized the database to improve query performance. Then, I wrote stored procedures and used indexing to optimize search efficiency. I implemented the recommendation logic using a combination of SQL joins, window functions, and a Python script to run batch computations asynchronously. Additionally, I used caching mechanisms like Redis to reduce redundant database queries.