◇ Could this help me?
Yes. The diagnosis is accurate and worth internalizing: adding a column per feature is exactly how a generated schema degrades, and normalization, indexes, and migrations are the right list of what is missing. Treat the specific column count as rhetorical rather than a threshold, since wide tables are sometimes a deliberate and correct choice, and the real signal is missing relationships and missing indexes rather than the number of columns alone. This is a general practice.
Review the database schema in this project and report on its design rather than its syntax. Identify any table accumulating unrelated columns, list the relationships that should exist but are not modeled, name every query that runs without an index to support it, and tell me whether migrations and a backup strategy exist. Then propose a normalized structure as a migration plan I can review, ordered so each step is safe to run on live data.