CV

Bachelor of Science

Columbia University

Computer Engineering

Bachelor of Art

Grinnell College

Computer Science

Fast Distribtued Selection with Graphics Processing Units

IEEE Open Journal of the Computer Society, Under Review

We introduce two completely parallel algorithms for exactly solving the distributed multiple order statistic problem. The main algorithm, DIBMS, keeps communication costs proportional to the number of order statistics with minimal communication of actual data. Moreover, the work done at the remote nodes is fully parallelized on graphics processing units.

Software Engineer

Digital Industry Software, Siemens AG

Software Engineer for NX, a leading CAD tool utilized by major companies such as SpaceX and Mercedes-Benz.

Enhancing the Underlying System of Large Language Models

Columbia University, advised by Professor Kostis Kaffes

Currently implementing intra-OP parallelism for the vLLM model to dynamically adjust GPU cores based on real-time memory and compute requirements.

• High-Performance Systems
• Large Language Model

Riker: Always-Correct and Fast Incremental Builds

Grinnell College, advised by Professor Charlie Curtsinger

Riker is a forward build system that guarantees fast, correct builds. Riker builds are easy to specify; in many cases a single command such as gcc *.c suffices. From these simple specifications, Riker automatically discovers fast incremental rebuild opportunities.

• Build Systems

Fast Distributed Selection with Graphics Processing Units

Grinnell College, advised by Professor Jeffrey Blanchard

Extracting multiple order statistics from a huge, distributed data set provides important information about the distribution of the values in the full set of data. We proposed an algorithm, DistributedIterativeBucketMultiSelect, or DIBMS, which selects the percentile from a data set of 268 million values distributed at eight locations in less than a quarter of a second.

• Distributed System
• Graphics Processing Unit
• Parallel Computing

Linux Kernel Development

Hack into the Linux Kernel to implement and modify core components like the scheduler, memory management, and file system.

• Operating System
• Linux Kernel