Chapter 7: Deploying Deep RL Agents to the Cloud
The cloud has become the de facto platform of deployment for AI-based products and solutions. Deep learning models running in the cloud are becoming increasingly common. The deployment of reinforcement learning-based agents to the cloud is, however, still very limited for a variety of reasons. This chapter contains recipes to equip yourself with tools and details to get ahead of the curve and build cloud-based Simulation-as-a-Service and Agent/Bot-as-a-Service applications using deep RL.
Specifically, the following recipes are discussed in this chapter:
- Implementing the RL agent’s runtime components
- Building RL environment simulators as a service
- Training RL agents using a remote simulator service
- Testing/evaluating RL agents
- Packaging RL agents for deployment – a trading bot
- Deploying RL agents to the cloud – a trading Bot-as-a-Service
