It is a reality of our contemporary society that data has grown exponentially, rendering fast search a dire necessity. The evolution of hardware and the increase of main memory capacity have led to reconsider the implementation of logical and relational operators on main memory databases. In this thesis we explore optimal search algorithms on sorted data, the inspiration of which came from the well-known paper ”The Case for Learned Index Structures” (maybe cite here), announced at the database community of Sigmod in 2018. Its main idea was that a machine learning model can be taught the sort order structure of lookup keys and utilize this signal to effectively predict the position or existence of records. Our approach was to build a similar model that can partition the sorted data and create buckets such that data will follow a uniform distribution with the aid of a V-Optimal histogram. Consequently, we can apply fast searching algorithms based on this distribution.