Georgios Liargkovas

Γιώργος Λιαργκόβας
Pronunciation: 'Yor-gos Li-ar-go-vas' - The 'G' in Georgios is pronounced like the 'Y' in 'Yes'.

profile pic
PhD Student @ Columbia
Member of:
Columbia Icon Nebula   Daplab Icon DAPLab
📍 DSI 425
📧 Email me
🔑 ssh public key
📄 Curriculum Vitae
About | Publications | Blog


About Me

I am a PhD student in the Computer Science Department at Columbia University, advised by Prof. Kostis Kaffes. My research integrates LLM agents with traditional machine learning to improve performance and efficiency in operating systems—with an emphasis on CPU schedulering, workload-aware adaptation, and safe, auditable control.

I’m part of Columbia’s Nebula and DAPLab groups. I hold a BSc from the Department of Management Science & Technology, Athens University of Economics and Business.

Previously, I was a research affiliate with the Atlas Systems Group at Brown University, advised by Prof. Nikos Vasilakis. I contributed to PaSh and also developed hs, an out-of-order shell script executor. We also built binpash - try stars - try , a lightweight shell-based sandbox.

Before Brown, I was a research associate at BALab with Prof. Diomidis Spinellis, where I worked on empirical software engineering topics.

Today my work centers on agents for system optimization (e.g., LLM-driven parameter tuning and control) that blend robust systems abstractions with data-driven adaptation.

Agentic Systems LLM Agents ML for Systems OS Tuning Scheduling Shell

News

Apr 2026
service
I will be serving on the Program Committee of AdaptFM, the Resource-Adaptive Foundation Model Inference workshop at ICML ‘26.
Feb 2026
milestone
I received my M.S. in Computer Science from Columbia CS.
Aug 2025
paper
Our paper, “Empowering machine-learning assisted kernel decisions with eBPFML,” will appear at eBPF ‘25!
Jul 2025
service
I will be serving on the Artifact Evaluation Committee for EuroSys 2026.
Jun 2025
award
Honored to receive a fellowship from the Gerondelis Foundation!
Jun 2025
project award
Our research proposal “From Noisy Signals to Clear Decisions: Optimizing Long-Term Outcomes with Relative Feedback and Causal Inference” will be funded by the Columbia–Dream Sports AI Innovation Center. Excited to pursue this work!
Mar 2025
service
I will be serving on the Artifact Evaluation Committee for USENIX ATC ‘25.
Nov 2023
service
I will be serving as Junior PC member for EuroSys 2024!
Oct 2023
talk
I joined the PaSh research workshop, for a week of PaSh hacking! I presented a deep dive of hs.
Jun 2023
project
binpash - try makes it to the top of Hacker News and surpasses 1000 GitHub stars! (currently stars - try forks - try )
Apr 2023
paper
Speculative shell paper was accepted to HotOS 23’!
Mar 2023
milestone
Graduated with Honors from my Bachelor’s Degree! 🎓

Select publications

  1. ICLR ’26
    Speculative Actions: A Lossless Framework for Faster AI Agents
    Naimeng Ye*, Arnav Ahuja*, Georgios Liargkovas*, Yunan Lu*, Kostis Kaffes, and Tianyi Peng
    In International Conference on Learning Representations (ICLR) ’26 2026
    🏆 Oral, To appear
    [pdf]
  2. OSDI ’26
    hS: Speculative Script Reordering at Subprocess Granularity
    Georgios Liargkovas*, Di Jin*, Tianyu (Ezri) Zhu, Dan Liu, A. Bolun Thompson, Anirudh Narsipur, Seong-Heon Jung, Siddhartha Prasad, Diomidis Spinellis, Michael Greenberg, Konstantinos Kallas, and Nikos Vasilakis
    In USENIX Symposium on Operating Systems Design and Implementation (OSDI ’26) 2026
    To appear
    [pdf]
  3. OSDI ’26
    Controlling Opaque-Component Effects with Semisolates and Ef
    Evangelos Lamprou, Tianyu (Ezri) Zhu, Di Jin, Grigoris Ntousakis, Georgios Liargkovas, Calvin Eng, Konstantinos Kallas, Michael Greenberg, and Nikos Vasilakis
    In USENIX Symposium on Operating Systems Design and Implementation (OSDI ’26) 2026
    To appear
    [pdf]
  4. ML4Sys ’25
    An Expert in Residence: LLM Agents for Always-On Operating System Tuning
    Georgios Liargkovas, Vahab Jabrayilov, Hubertus Franke, and Kostis Kaffes
    In ML for Systems Workshop at NeurIPS 2025
    [pdf]
  5. PACMI ’25
    Set It and Forget It: Zero-Mod ML Magic for Linux Tuning
    Georgios Liargkovas, Prabhpreet Singh Sodhi, and Kostis Kaffes
    In Proceedings of the 2025 ACM Workshop on Practical Adoption Challenges of ML for Systems (PACMI ’25) 2025
    [pdf]

Personal

In my free time, I enjoy a variety of indoor and outdoor activities. Long-distance running is close to my heart (feel free to follow me on Strava), either solo or with the company of others. I also love cooking, often trying and adding personal twists to recipes from many cuisines. Hopefully, I'll share some of my recipes here soon. Additionally, I'm passionate about curating playlists. Alternative Rock, Indie, and Jazz are among my top genres. In earlier days, I used to play the piano and take music theory lessons.