Vignesh Kothapalli

vk_prof_pic.jpeg

Hi :wave:, I’m a Senior Software Engineer, ML in the Foundation Models and Data team at LinkedIn AI. My current research focuses on post-training techniques for 360Brew, which is LinkedIn’s foundation model for recommendation.

Previously, I worked with Prof. Joan Bruna (NYU Courant), Prof. Yaoqing Yang (Dartmouth) and Prof. Rahul Mazumder (MIT) on topics related to the learning dynamics of neural networks. I also spent time at IBM as a TensorFlow contributor and TensorFlow-IO maintainer. I did my MSc in Computer Science at NYU Courant, and B.Tech in Electronics and Communication Engineering from IIT Guwahati.

Research

I’m primarily interested in understanding the computational aspects of learning in neural networks, and using these insights to develop novel modeling techniques.

  • Foundation Models: Chain-of-Thought reasoning, In-Context Learning and Multi-Task Learning in Transformer based LLMs.
  • Learning Dynamics: Understanding the interplay between feature evolution, weight matrix spectra and generalization in shallow and deep neural networks.
  • Efficient Training/Inference: Developing low-level (randomized) algorithms for training and inference of large scale LLMs/GNNs.

Whats New

Academic Service

  • Conference Reviewer: NeurIPS 2024, ICML 2025
  • Journal Reviewer: : IEEE Transactions on Cybernetics, IEEE Access, IEEE Transactions on Industrial Informatics, TMLR

Selected Publications

See full list at Google Scholar.
  1. arxiv
    liger.gif
    Liger Kernel: Efficient Triton Kernels for LLM Training
    Pin-Lun Hsu, Yun Dai, Vignesh Kothapalli, and 7 more authors
    arXiv preprint arXiv:2410.10989, 2024
  2. NeurIPS
    shallow_nc.png
    Toward Understanding How the Data Affects Neural Collapse: A Kernel-Based Approach
    Vignesh Kothapalli, and Tom Tirer
    In NeurIPS 2024 Workshop on Symmetry and Geometry in Neural Representations , 2024
  3. NeurIPS
    graphnc.gif
    A Neural Collapse Perspective on Feature Evolution in Graph Neural Networks
    Vignesh Kothapalli, Tom Tirer, and Joan Bruna
    Advances in Neural Information Processing Systems, 2023
  4. ICML
    rlap.png
    Randomized Schur Complement Views for Graph Contrastive Learning
    Vignesh Kothapalli
    In International Conference on Machine Learning , 2023
  5. TMLR
    nc_review.png
    Neural Collapse: A Review on Modelling Principles and Generalization
    Vignesh Kothapalli
    Transactions on Machine Learning Research, 2023