HomeResearchProjectsCV

Research

My research sits at the boundary between Kubernetes and the GPUs underneath it, and runs along two threads. The first is sharing and scheduling: how a single device is divided between workloads, and how those workloads are placed across heterogeneous hardware, so that scarce GPU capacity is used efficiently rather than left idle by whoever claimed it first. The second is observability: hardware monitoring of the devices themselves, alongside profiling of the work running on them, so that what happens on a GPU can be attributed back to the workload responsible for it. The threads meet in practice: dividing a GPU well depends on being able to see how it is actually being used.

Google Scholar

Publications

  1. Synchronous Distributed Training with Runtime-Adaptive Mechanisms in Hybrid Cloud Environments

    T. A. Vuong, T. L. Hoang, H. Le, Y. H. Kim

    IEEE Access · 2026

  2. A Design for Continuous Low-Overhead CUDA-API Observability in Kubernetes GPU-Sharing Systems

    T. L. Hoang, Y. H. Kim

    KICS Summer Conference · 2026

  3. A Design of Cloud-based Personal Machine Learning Workspace

    T. L. Hoang, T. N. D. Thong, Y. H. Kim

    KICS Summer Conference · 2025

  4. A Comparison of Two SDN Controllers Integration in OpenStack Environment

    T. L. Hoang, D. Q. Pham, T. T. M. Dinh

    International Conference on Intelligent Systems & Networks (ICISN) · 2025

  5. Performance Evaluation of OpenStack Cloud Integrated with SDN Controller

    M. T. T. Dinh, T. L. Hoang, Q. D. Pham, D. N. Nguyen

    REV Journal on Electronics & Communications · 2024

  6. Evaluation of Network Performance between Virtual Machines on Cloud Computing integrating SDN Controllers on Virtual Environments (in Vietnamese)

    T. L. Hoang, Q. D. Le, T. P. D. Chu, T. T. M. Dinh

    REV-ECIT · 2023