cv

Backend Engineer

Standard Practice AI

Building AI for outbound calls from medical practices.

Scientific Computing Intern

Flatiron Institute, Simons Foundation

This position was at the Flatiron Institute's Scientific Computing Core. I implemented and benchmarked parallel high-performance large matrix decompositions using CUDA and SYCL, a cross-platform GPU programming language.

• Improved LDLT factorization performance with high-performance GPU computing by up to 150x over CPUs
• designed a research poster based on my work
• communicated results to interdisciplinary computational scientific teams at an internal conference

Undergraduate Research Assistant

Geometric Computing Lab, New York University

I was an undergraduate research assistant at NYU's Geometric Computing Lab, a physical simulation and geometry processing research group. I worked on multiple research projects here as an undergraduate.

• Maintained the CMake build pipeline and re-added CUDA support for SymbolicLib, an expression tree compiler for optimizing repeated sparse matrix operations
• Integrated cuSolverDN, a GPU-based dense linear solver written in CUDA, into polyfem, a C++ finite element library, and polysolve, a wrapper for linear solvers
• Compared assembly and solving times for finite element method problems between dense and sparse matrix data structures using matplotlib, numpy, CuPy, PolyFEM, libigl, and polysolve
• assembled a research poster based on this project, presented work at the NYU Undergraduate Research Conference to a general audience

Tutor for CSCI-UA 102 Data Structures, Course Assistant for CSCI-UA 102 and CSCI-UA 201

Courant Institute of Mathematical Sciences, New York University

I was asked to be a tutor and course assistant for two computer science classes based on my strong performance as a freshman in computer science.

• Scheduled regular tutor office hours for intro-level data structures and systems programming students learning Java, C, and x86 assembly, for a combined total of over 100 students during the summer
• Tracked attendance across both courses with Zoom and Microsoft Excel
• Spearheaded four virtual recitation sections over Zoom for overseas students to encourage attendance and provide extra support during COVID-19

Grader for CSCI-UA 102 Data Structures

Courant Institute of Mathematical Sciences, New York University

I was asked to be a grader for an introductory data structures class based on my strong performance during the semester. I held this position for my first two years at NYU.

• Graded weekly homeworks/assignments for an intro-level data structures course taught in Java using Gradescope and JUnit
• Provided thorough feedback to a section of 25 students in a total course size of over 100
• Connected with instructors to proofread and verify assignments and other materials prior to distribution to students

Counselor

Herricks Summer Music Camp

Volunteer counselor for a high school summer music camp.

• Participated in demonstration ensembles with fellow counselors
• Engaged with brass students in the 3rd-8th grades to develop musical skills
• Selected, arranged, and composed classical and jazz repertoire for young student ensembles

Bachelor of Arts

New York University

Computer Science and Mathematics

• Computer Graphics (Graduate)
• Computer Vision (Graduate)
• Virtual Reality
• Machine Learning
• Intro to Cryptography
• Quantum Computing
• Honors Numerical Analysis

Dean's Undergraduate Research Fund Grant

New York University

The Dean's Undergraduate Research Fund is a $2,000 grant awarded to undergraduate students showing promise in research by submitting an abstract and detailed research plan. All students receiving this award present at NYU's Undergraduate Research Conference.

Scaling the Finite Element Method to Small Problems with Dense Linear Solvers

Inquiry (NYU Undergraduate Research Journal)

I used dense matrix data structures for small Laplacian problems with the finite element method. They may be faster if the problem is small, and the memory is already in cache and already zeroed. This was presented at NYU's undergraduate research conference and published in Inquiry, NYU's undergraduate research journal.

Implementing and Benchmarking a Parallel LDLT Factorization with SYCL

Simons Foundation Internal Conference

I compared the speed of SYCL, a cross-platform GPU language, versus CUDA and CPU programming, with a highly-parallel block matrix factorization. This was presented at an internal conference for interns at the Flatiron Institute.

LDLT Benchmark

An implementation of the LAPACK routines sytrf() and ssptrf() in SYCL, a cross-platform GPU programming language.

Dense Assembly Timing

I measured times for using dense versus sparse matrices in finite element assembly and solving.

Polysolve (Contributor)

I updated a linear solver interface to include cuSolverDN, a GPU-based dense linear solver written in CUDA.

Quod Erat Osthenes

A multi-threaded prime number finder in C using POSIX threads.