Conventional grids are defined by restricted communication capabilities, one-way electricity transmission, and centralised generation. Consequently, issues such as unstable voltage, insufficient load balancing, and increased energy losses occur. Unlike a traditional grid, a smart grid is a modernised electrical network that is more efficient, reliable, and sustainable by utilising digital technologies and two-way communication. The suggested study combines machine learning (ML)-based methods with proportional integral derivative (PID) controllers for sustainable energy management in order to maximise performance. While PID controllers help to reduce the volatility of renewable energy sources like solar and wind in order to maintain stability in voltage, ML algorithms-including neural networks-predict changes in energy demand and supply. The MLPID hybrid approach reduces energy losses, enhances response time, boosts grid durability, and aids in load distribution.