About

I’m a software engineer at Meta, where I work on infrastructure for running large language models over large volumes of data. Before that I spent about 20 years building systems that move a lot of data reliably, including Amazon Data Firehose at AWS, real-time bidding at Amobee, and data platforms at IBM.

I’m a Fellow of the BCS (The Chartered Institute for IT) and maintain steadybatch, an open-source library and benchmark for reliable batch LLM inference.

Posts here are my own and don’t describe my employer’s systems.