Skip to content
View kevinknights29's full-sized avatar
🚀
🚀

Block or report kevinknights29

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
kevinknights29/README.md

Kevin Knights

Computational Mathematics & Scientific Computing

I build high-performance numerical software for mathematically demanding problems.

I am completing an MSc in High-Performance Computing (Computational Mathematics) at Trinity College Dublin’s School of Mathematics. My current interests include numerical linear algebra, sparse linear system solvers, Krylov subspace methods, and GPU-accelerated scientific computing.

Before specializing in computational mathematics, I spent five years building production machine learning and data systems at Dell Technologies and AES. That experience shaped how I approach research software: mathematical ideas should lead to efficient, reliable implementations that work at realistic scale.

Research

My MSc thesis, supervised by Prof. Kirk M. Soodhalter, investigates when communication-avoiding Krylov methods can outperform conventional approaches in distributed multi-GPU environments.

The principal contribution is a dimensionless regime map that predicts whether communication avoidance should pay on a given machine using problem parameters and independently calibrated machine constants—before running the full implementation.

The model correctly predicted:

  • No benefit on a single-NUMA-node CPU
  • A benefit across four NVIDIA V100 GPUs on two nodes
  • A 1.78–1.80x speedup over an equal-work control

Research interests

  • Numerical linear algebra
  • Sparse linear system solvers
  • Krylov subspace methods
  • Matrix functions and exponential integrators
  • GPU and multi-GPU computing
  • Communication-avoiding algorithms
  • Performance modeling and optimization
  • Numerical methods for computational finance
  • Efficient training and inference of machine learning models

Technical focus

  • Languages: Python and C++
  • Parallel computing: CUDA, MPI, OpenMP, and NCCL
  • Scientific computing: numerical simulation, sparse linear algebra, and performance analysis
  • Machine learning: production ML systems, model deployment, distributed data processing, and inference optimization

Connect

I am interested in research and engineering conversations at the intersection of mathematics, high-performance computing, and machine learning.

LinkedIn Email


"What is the fruit of these teachings? Only the most beautiful and proper harvest of the truly educated——tranquillity, fearlessness, and freedom. We should not trust the masses who say only the free can be educated, but rather the lovers of wisdom who say that only the educated are free." ~ Epictetus.

Pinned Loading

  1. GenAI_Papers GenAI_Papers Public

    This project contains a list of interesting research papers in the field of GenAI

    70 19

  2. caksm caksm Public

    Communication-Avoiding Krylov Exponential Integration for Multi-GPU Option Pricing

    C++