Laxminarayana Vadnala

Lax

Ph.D in Computer Science

University of Notre Dame, South Bend, IN

lvadnala@nd.edu

laxminarayana.vadnala1997@gmail.com

View Resume

Bio

I am a Second-year Ph.D. student in Computer Science at the University of Notre Dame, where my research focuses on scalable systems, High-Performance Computing (HPC), distributed systems, Linux systems, and infrastructure for both Scientific and AI workloads. I am currently resarching at the Cooperative Computing Lab , advised by Prof. Dr. Douglas Thain.

A major direction of my current research is distributed GPU computing and resource management for AI workloads. I am working on a user-space distributed GPU execution and resource-management system that explores mechanisms similar in spirit to technologies such as NVIDIA MPS, while extending the idea toward distributed and high-throughput computing environments. The goal is to integrate this work with TaskVine, enabling researchers and practitioners to more efficiently utilize GPU resources for workloads such as LLM training, inference, and other GPU-intensive AI applications.

This work currently focuses on Linux and NVIDIA GPUs, with an emphasis on understanding and managing GPU resources from user space while interacting closely with operating-system mechanisms and the underlying GPU runtime. In the longer term, I am interested in extending these ideas beyond a single GPU vendor and investigating runtime capabilities inspired by modern AI-serving and inference systems such as vLLM and llm-d. Areas I am exploring include efficient GPU sharing, workload scheduling, distributed inference, resource isolation, dynamic allocation, batching, and improving GPU utilization across collections of AI tasks.

Another major part of my research involves building and studying large-scale simulation and computing systems, particularly through the Safety-Aware Drone Ecosystem (SADE) project. SADE combines multiple simulation and autonomy components to support large-scale sUAS experiments, and my work focuses on improving the scalability, reliability, observability, and resource efficiency of the overall system.

A growing part of this work focuses on Linux systems and kernel internals. I work with mechanisms and interfaces such as ProcFS (/proc), Control Groups (/sys/fs/cgroup), process hierarchies, memory accounting, CPU utilization, container isolation, namespaces, and GPU resource monitoring to understand how complex applications consume system resources at runtime. I have been developing lightweight observability and systems-analysis tools in Rust that reconstruct process hierarchies, correlate processes with containers, and track CPU, memory, and GPU behavior throughout the lifetime of distributed workloads.

As part of my upcoming work on SADE, I am investigating how information exposed through ProcFS (/proc) and other Linux interfaces can be used to reconstruct and visualize the complete lifecycle of active processes. This includes understanding parent-child process relationships, container membership, memory consumption, CPU behavior, and GPU utilization as workloads evolve. The broader goal is to make the internal behavior of large distributed applications easier to visualize, diagnose, and optimize as they scale.

I also began my journey as a Linux kernel contributor. Through this effort, I hope to develop a deeper understanding of kernel subsystems and contribute upstream over time. I am particularly interested in running the mainline linux kernel on the variuos used android mobiles and build a cluster by connecting them as mesh nodes. Areas related to process management, resource control, scheduling, memory management, cgroups, observability, performance, and interactions between the Linux kernel and accelerator-based workloads.

More broadly, my research is driven by a systems question: how can we make large-scale computing resources easier to share, observe, and utilize efficiently? I am particularly interested in problems at the intersection of HPC and AI infrastructure, including how clusters containing CPUs and GPUs can support increasingly dynamic AI and scientific workloads without sacrificing utilization, isolation, performance, or scalability.

Prior to my Ph.D., I worked in industry for more than four years, including roles as a Solution Architect at Southwest Airlines and a Senior Software Engineer at Qualcomm, along with other software engineering roles. My industry experience, combined with my current systems research, has shaped my interest in building computing systems that are not only scalable in principle but also practical, observable, and reliable in real-world environments.

My broader research interests include:

  1. Distributed GPU Computing and GPU Resource Management
  2. Systems for AI, LLM Training, and Inference
  3. High-Performance and High-Throughput Computing
  4. Distributed and Scalable Systems
  5. Linux Kernel and Operating Systems
  6. Performance Engineering and Systems Observability
  7. GPU Scheduling, Sharing, and Heterogeneous Computing
  8. Container and Resource Management
  9. Large-Scale Simulation Systems

I am actively seeking Research internships for 2027 at national laboratories and research-oriented technology companies, particularly opportunities involving HPC, distributed systems, GPU systems, Linux/operating systems, AI infrastructure, LLM systems, performance engineering, or large-scale computing. I am especially interested in research problems involving the design of systems that allow AI and scientific workloads to efficiently utilize distributed CPU and GPU resources at scale.

Education

Univeristy of Notre DameAug. 2025 - Present

Ph.D. in Computer Science

High-Performance Computing

Saint Louis UniversityAug. 2023 - May 2025

M.Sc. in Computer Science

News

July 6, 2026 Talk
Presented our work on SADE-SIM: A Scalable Simulation Platform for Validating City-Scale Multi-sUAS Missions at SE4ES 2026 in Montreal, Canada
April 2, 2026 Publication
paper to 2nd International Workshop on Software Engineering for Engineering Simulations, and Simulation Engineering for Engineering Software (SE4ES) 2026 with title SADE-SIM: A Scalable Simulation Platform for Validating City-Scale Multi-sUAS Missions got accepted
March 24, 2025 Career
Got Admission confirmation for Ph.D at University of Notre Dame in Department of Computer Science will start working with Professor Douglas Thain in stream of High-Perfromance Computing.
August 19, 2024 Achievement
As part of sUAS—Human-Centered Computing Lab (sUAS HCC) in collaboration with Dr. Douglas Thain, from Cooperative Computing Lab at University of Notre Dame, we successfully deployed the SADE-SIM frontend platform
July 2, 2024 Career
Joined sUAS—Human-Centered Computing Lab (sUAS HCC) under Dr. Ankit Agarwal as Graduate Research Assistant at SLU.
August 21, 2023 Career
Started at SLU as M.Sc. in Computer Science Student.

Projects

Safety Aware Drone Ecosystem's Back end and Front end (SADE)

SLU Graduate Research Assistant

(July 2024 - Present)

Safety Aware Drone Ecosystem's Back end and Front end (SADE) Description

  • Implemented the whole back end from scratch with MQTT and Database integration to facilitate the transfer of drone configuration configured by user to SADE simulation stack which comprises of PX4, Gazebo and Unreal Game Engine to enable simulation.
  • Optimized the existing front end to support real time changes from user for butter smooth experience. Also integrated Cesium JS components to enable Google Maps 3D tiles for precise location configuration for better user experience.
  • Worked with Tech Stack Python, Fast API, Mosquitto MQTT, Mongo DB, Docker, GitHub Actions, React JS, Cesium JS, Docker Compose
Safety Aware Drone Ecosystem (SADE) Simulation Plugins

SLU Graduate Research Assistant

(July 2024 - Present)

Safety Aware Drone Ecosystem's Simulation Plugins

  • Developed and Managing the Gazebo physics engine plugins like Real time Terrain Loaders when drone fly using the terrain information from the 3D tiles of Cesium for simulating precise taking off, landing, obstacles for real time collision and more physics operation
  • Improvised the existing wind plugin from PX4 Gazebo as per the SADE project requirements in real time while applying the continuous wind force on the simulated drones in simulation
  • Developed a Drone Pose sender (Gazebo Plugin) and Receiver (CLI tool) which basically sends the Pose of Drone from gazebo in real time as UDP packets and Pose Receiver will be receiving the UDP packets in real time. This includes real time Endian encoding and decoding
  • Worked with Tech Stack C++ 11, SDF, Gazebo 11 C++ library, ROS Noetic, Linux Bash Scripts
Safety Aware Drone Ecosystem (SADE) Workflow Management

SLU Graduate Research Assistant

(July 2024 - Present)

Safety Aware Drone Ecosystem's Workflow Management

  • Developed the workflow management system for integrating and scaling of Unreal Engine Pods, Unreal Pixel Streaming Pods, GUI and Back end Pods, PX4 Auto Pilot Pods and Gazebo Pods
  • Unreal Engine Pods share the GPU resources, we achieved this using NVIDIA Container toolkit, and streaming the display over network via Pixel Streaming to Front end
  • Worked with Tech Stack Python, Shell Script, Docker, Docker compose
Drone Response (DR) Live Camera feed pipeline

SLU Graduate Research Assistant

(July 2024 - Present)

Safety Aware Drone Ecosystem's Workflow Management

  • Implemented the whole back end from scratch with MQTT and Database integration to facilitate the transfer of drone configuration configured by user to SADE simulation stack which comprises of PX4, Gazebo and Unreal Game Engine to enable simulation
  • Optimized the existing front end to support real time changes from user for butter smooth experience. Also integrated Cesium JS components to enable Google Maps 3D tiles for precise location configuration for better user experience
  • Worked with Tech Stack Python, Fast API, Mosquitto MQTT, Mongo DB, Docker, GitHub Actions, React JS, Cesium JS, Docker Compose

Publications

Most recent publications on Google Scholar.

SADE-SIM: A Scalable Simulation Platform for Validating City-Scale Multi-sUAS Missions

FALaxminarayana Vadnala, Lucas Parzianello, Bohan Zhang, Michael Murphy, Ankit Agrawal, Douglas Thain

FSE Companion '26: Proceedings of the 34th ACM International Conference on the Foundations of Software Engineering (SE4ES 2026)

Paper doi: https://doi.org/10.1145/3803437.3805540

SADE-SIM: A Scalable Simulation Platform for Validating City-Scale Multi-sUAS Missions

FALaxminarayana Vadnala, Lucas Parzianello, Bohan Zhang, Michael Murphy, Ankit Agrawal, Douglas Thain

FSE Companion '26: Proceedings of the 34th ACM International Conference on the Foundations of Software Engineering (SE4ES 2026)

Paper

FA indicates first Author contribution.
EQ indicates equal contribution.

Talks

SADE-SIM: A Scalable Simulation Platform for Validating City-Scale Multi-sUAS Missions

ConferenceMontreal, Canada (2026)

Autonomous deployment using Github Actions (CI/CD) pipeline setup and best practices (Guest lecture) at Saint Louis University

Guest LectureSaint Louis, MO, USA (2024)

Mentee

Prof. Douglas Thain (University of Notre Dame), Aug 2025 - Present
Dr. Ankit Agarwal (Saint Louis University), Jul 2024 - Aug 2025

Vitæ

Full CV in PDF.

Collaboration Graph

This graph shows my collaborators, advisors, institutions, research labs, and major projects. Drag nodes to rearrange the graph. Hover over a node or connection for details.

Skills Graph

This graph shows my Technical Skills, Expertise with tools and technologies. Drag nodes to rearrange the graph. Hover over a node or connection for details.

Laxminarayana Vadnala
lvadnala@nd.edu