Sunnyvale, California, United States
Building out the Inference SW team at Etched - we're hiring! I'm excited to tackle the problems required to scale out AI. If you feel the same and enjoy working on ML systems, please reach out =)
Building out a Inference SW team at Etched - we're hiring! Our SW stack is responsible for programming the chip and efficient serving across multiple chips and systems. I'm personally working on collectives. Folks in our group work on problems like our programming model, optimal batch scheduling, our distributed runtime and mapping models to our HW (analagous to GPU kernel programming). It's a ton of fun =)
Senior software engineering manager at Google Platforms, which is the org responsible for new systems (e.g. accelerators, servers, storage, networking, etc.) developed and deployed at Google. We develop all kinds of crazy systems and my team helped make them work at scale. I built and lead a team of ~50 called System Modeling. This is primarily a collection of infrastructure projects that help manage these systems at scale. As the name implies, there is a heavy focus on data infrastructure. Projects include telemetry agents, fleet-wide detection, graph-based configuration service, health database, etc. Prior Google projects include: -simulation/modeling infrastructure to rapidly develop SW for cutting edge chips (e.g. TPU) and systems (e.g. smart NICs, accelerator trays, dis-aggregated storage) -a HW accelerator project for one of Google's billion user services
Altera acquired by Intel. Built and managed a 15-person SW team building out a FPGA acceleration stack. Chiefly responsible for development but also did customer engagement.
Founding member of Altera Virtualization Lab, which focused on generalized FPGA acceleration. Worked on live migration of VMs with attached FPGA accelerators, which yielded some patents. Before that, worked on toolchains (debuggers, linkers, co-sim, etc.).