Aubrey Clark

I build machine learning systems. Trained as an economist, I now work on infrastructure, optimization, and adversarial ML.

I'm a data scientist and technical lead at Google, working on predictive maintenance, capacity planning, and congestion control for Google's global network infrastructure. Previously I was at Twitter (serving stack optimization, spam detection) and Wealthfront (automated financial planning). I did my PhD in economics at Harvard, where I worked on mechanism design with Eric Maskin.

CV | GitHub | Google Scholar

PROJECTS

Communication Systems from Scratch
BGP, TCP/IP, Audio Modem, QUIC, BBR, Protocol Buffers, HTTP/3, and gRPC — all from scratch.

Probabilistic Serial Mechanism
Fair allocation algorithm implementation.

Constrained Birkhoff-von Neumann
Budish-Che-Kojima-Milgrom algorithm for course allocation.

RESEARCH

Contracts for Acquiring Information (with G. Reggiani), 2017

Capacity Constraints in Principal-Agent Problems, 2017

Core Equivalence with Large Agents, 2017

Nostr: npub1t6gu5asl2636txpyeuqnwac8y7zc5035mjae2qdjq48z6q8p0mhqteajea