can partial upsert be adapted in schema evolution throughout realtime data?

Could you elaborate what you mean? Are you asking if partial upsert continues to work if you add more columns to your schema?