FedPay: An Incentive-Based Protocol For Scalable Federated Learning

  • Position: Machine Learning Research Engineer
  • Type: Part-Time
  • Duration: April 2020 - Present
  • Supervisor: Dr. Sujit Gujar
  • Laboratory: Machine Learning Lab
  • Location: International Institute of Information Technology, Hyderabad, India
  • Visit: https://mll.iiit.ac.in

Working as a research lead and sole engineer to develop a fault-tolerant, incentives-based, system design protocol for federated learning at scale. Proposed and formulated novel agent sampling and incentive technique to solve problems like pace steering, reduced training time, maximum client commitment, honest contributions, maximum system utilization, and crowd training. Built a scalable federated learning framework from scratch using PyTorch & PySyft.