Jinyang Li 李金扬 Professor Department of Computer Science Courant Institute of Mathematical Sciences 60 5th Ave New York, NY 10011 Contact: firstname at cs nyu edu

The goal of my research is to build easy-to-use and high-performance distributed infrastructure to simplify application development.

We live in an era with unprecedented amounts of data. Popular web applications store the data of billions of users and process these data offline to gather business insights. The increasing deployment of networked sensors, cameras, monitors, and devices also generate massive amounts of data. How to store such vast amounts of data efficiently and reliably across many machines? How to let applications query and update distributed data safely and quickly in the face of concurrency and failure? How to run state-of-the-art algorithms to analyze data at scale so that we can make sense of the collected data in real time? These are the principal questions that my research group aims to address.

I work with a group of talented students in the NYU systems group.

Current Projects

• Deep Graph Library is a framework for doing deep learning on graphs
• Tofu, an automatic parallelization framework for deep learning systems

Some older projects:

• Spartan, a system for programming with distributed arrays (aka distributed NumPy).
• Janus, RoCoCo and IC3 can commit transactions in a distributed or multicore setting more efficiently and minimize aborts.
• Pilaf and Cell are two super-fast in-memory storage systems based on RDMA.
• Piccolo, a programming paradigm for writing distributed in-memory computations.
• SumUp, Sybil-resilient content voting.
• WheelFS, a wide-area file system.