This document summarizes a student's seminar presentation on using deep reinforcement learning for intelligent traffic light control. It notes that India loses $22 billion annually to traffic congestion, which causes economic losses and increased accidents. As populations grow, transportation demands increase traffic issues. The student proposes using deep reinforcement learning to dynamically adjust traffic light durations based on real-time road conditions from cameras, with the goal of minimizing wait times at intersections. The model would use number of cars at each direction as the state, light duration changes as actions, and cars passing per cars waiting as the reward to optimize via Q-learning.