I did my PhD at the Technion on Byzantine agreement, a postdoc at Cornell on BFT systems, fair ordering and metastability, and then worked on performance and fault tolerance for a sharded blockchain. I like problems where theory and practice keep each other honest. These days I'm a research scientist at Chainalysis.
See what I'm intoI'm a distributed-systems researcher , a traveler , a reader & a jiu jitsu enthusiast
Byzantine agreement: getting machines that don't trust each other to agree, with far less talking. My PhD gave the first sub-quadratic asynchronous algorithm.
Metastability: systems that look fine until one push sends them into a bad state they can't climb out of alone. Can we see it coming?
Who goes first? What it should mean for a replicated system to order requests fairly, and how to guarantee it.
Taking ideas off the whiteboard: fast BFT systems, and performance work at scale. Profiling, benchmarking, and hunting the real bottleneck.
Agreement without clocks used to cost a quadratic number of messages. We showed it can be done with nearly linear communication.
Read it → Watch the talk →What does it even mean for a shared object to be "correct" when some of the processes using it are lying? We defined it, and mapped what can be built from plain registers.
Read it → Watch the talk →Agreement that pays for the failures that actually happen, not the worst case you have to plan for.
Read it →Every part of a system can look healthy on its own, and the whole thing can still tip over. Can we predict it from measuring the parts?
Coming soon
