Really enjoyed this, Sonal. The point around rewards shaping behavior is so important, especially because in RL the challenge is not just making the system learn, but making sure it learns the right thing. This was a crisp and helpful explanation.