class KayongoBrian(Agent):
"""Machine learning engineer. Trains things until they work."""
observation_space = ["computer vision", "deep RL", "real-world data"]
action_space = ["train", "ship", "teach", "repeat"]
reward = "models that actually make it to production"
def policy(self, state):
return max(self.ideas, key=lambda a: a.q_value) # mostly greedy
while alive: explore(); exploit(); log_everything()