PRANAV BANAVATH completed Fundamentals of Reinforcement Learning: Racing Challenge

Built a complete DeepRacer reward-design dossier with iterative training evidence, exploit diagnosis, and a defended final shaping strategy. The candidate demonstrated leakage-aware reward design, preserved a speed term while conditioning it to avoid gaming, and documented how the reward behaved on both practice and unseen circuits.

• Leakage-Free Reward Shaping
• Measurement-Based Iteration
• Reward-Hack Diagnosis
• Generalisation Reasoning