About

My research has focused on the use of erasure coding for fault tolerance in distributed computing. What separates my research from traditional methods in the field is that I have focused on encoding the computations themselves as opposed to simply applying erasure coding to data. In particular, my research focuses on performing evaluating functions on the encoded data itself as opposed to decoding, computing, and re-encoding the data. This novel and emergent field of research has much potential to impact current distributed systems architecture since much of modern distributed computing involves efficient big data storage and analysis. Furthermore, these methods allow for efficient and theoretically optimal fault-tolerant computation on arbitrarily large distributed data-sets. The main types of computations I have focused on are matrix multiplication, regression, and feed-forward neural networks.