“The PPO (Proximal Policy Optimization) algorithm, developed by John Schulman in 2017, is considered the foundational predecessor to modern reinforcement learning techniques used for LLMs.”