Alexey Tumanov

Alexey Tumanov is an Assistant Professor in the School of Computer Science at Georgia Institute of Technology, part of the College of Computing. His research focuses on systems for machine learning, resource management, and scheduling in distributed environments. He holds a PhD from Carnegie Mellon University and conducted postdoctoral research at UC Berkeley under Ion Stoica. Previously, he worked at the University of Toronto and in industry on cloud computing and datacenter systems.

Education:

• PhD in Computer Science, Carnegie Mellon University (2019)
• Postdoc, UC Berkeley RISELab (2019)
• MSc in Computer Science, University of Toronto (2012)
• BSc in Computer Science, University of Toronto (2010)

Research Interests: Tumanov's work addresses challenges in distributed machine learning, including efficient resource management for inference and training pipelines, scheduling algorithms for soft-real-time systems, and federated learning. His lab (SAIL) develops systems like SARATHI-SERVE and RocketKV to optimize LLM inference and co-scheduling. He emphasizes practical deployments in healthcare (HOLMES) and cloud environments.

Key Contributions:

• Ray distributed framework for AI (OSDI 2018)
• ESCHER scheduler for ephemeral cloud resources (SoCC 2022)
• InferLine ML pipeline orchestration (SoCC 2020)

Awards:

• NSERC CGS-D3 Scholarship (2016)
• Best Paper at EuroSys 2016 (TetriSched)
• Google PhD Fellowship Nominee (2018)

Lab & Teaching: Leads the Systems for Artificial Intelligence Lab (SAIL) at Georgia Tech. Teaches advanced courses on operating systems (CS3210) and systems for ML (CS8803-SMR). Supervises 15+ graduate students across PhD and MS programs.