Jovan Stojkovic

Jovan Stojkovic

Incoming Assistant Professor

I will join UT Austin CS as an Assistant Professor in Fall 2026, and I am looking for prospective PhD students.

I am Jovan Stojkovic, an incoming Assistant Professor at the Department of Computer Science at the University of Texas at Austin. Before joining UT Austin, I will spend a year at Meta working with the AI and Systems Co-design group. I have completed my PhD at the University of Illinois at Urbana-Champaign, where I was advised by Professor Josep Torrellas.

My research focuses on cloud computing and datacenters, with two primary domains:

  1. ☁️ Cloud-native workloads, such as microservices and serverless computing
  2. 🤖 Machine Learning (ML) inference, particularly large language models (LLMs)

Cloud-native workloads represent a shift in how services are deployed and managed, offering scalability, a simplified programming model, and cost-effective billing. However, these benefits come with new challenges. Unlike traditional monolithic applications, microservices and serverless functions exhibit high communication overhead, fine-grained execution, bursty invocation patterns, and focus on optimizing for tail latency—all of which lead to poor performance, resource, and energy inefficiency on conventional cloud infrastructure. As these deployment paradigms gain traction, my research aims to co-design novel hardware platforms and software systems that deliver orders-of-magnitude improvements in performance, energy efficiency, and resource utilization for these emerging workloads.

In parallel, I focus on the rapidly growing demands of ML inference, especially the inference of generative LLMs. These models now serve latency-sensitive applications under strict Service Level Objectives (SLOs), often relying on power-hungry GPUs to meet performance demands. This has led to a dramatic rise in power draw and energy consumption from inference clusters. My work addresses this challenge through smart scheduling, workload placement, and system-level configuration tuning to reduce energy, power, and thermal overheads—while maintaining performance and accuracy guarantees.

Note to prospective students

🏆 Selected Honors and Awards

I completed my undergraduate studies at School of Electrical Engineering, University of Belgrade, Serbia. Every year (2017-2020), I was recognized as the best student of Computer Engineering and Information Theory Department.