Devcon VI

Reinforcement Learning for Query Pricing in The Graph
10-12, 10:00–10:30 (America/Bogota), Talk 3

Indexers​​ in The Graph protocol use a DSL called Agora​ to map query shapes to prices. However, manually populating and updating Agora models for each query is a tedious task, and, as a consequence, most indexers default to a flat pricing model. We have created and deployed reinforcement learning agents for Indexers to automatically compete on pricing. This talk will focus on our development process and our study of the market effects of multiple competing pricing agents.

I am an applied research scientist, able to drive research into novel directions.

With over 20 years in academia and industry, I have developed strong analytical & problem solving skills along with deep expertise in software engineering, robotics, visual perception and machine learning.

I am a fan of robotics, AI, design patterns and MOOCs.