Hossein Hosseini

Ph.D. Candidate in Computer Science | The University of Melbourne

prof_pic.jpg

Melbourne, Australia

hossein.hosseini@student.unimelb.edu.au

I am a Ph.D. Candidate in Computer Science at The University of Melbourne, working under the supervision of A/Prof. Adel N. Toosi and Prof. Chris Leckie. My research focuses on cost-efficient inference for large language models, exploring scalable deployment strategies, latency-reliability trade-offs, and introspective models that can self-evaluate their output quality.

My research interests include:

  • Large Language Models and efficient inference
  • LLM routing and model selection
  • Self-evaluating and introspective language models
  • Model compression and adaptation (e.g., LoRA)
  • Systems for Machine Learning
  • Cyber-Physical Systems and real-time embedded systems

I hold an M.Sc. in Computer Engineering from Sharif University of Technology (GPA: 19.41/20.0) and a B.Sc. in Computer Engineering from the same institution (GPA: 16.32/20.0). Before pursuing my Ph.D., I conducted research on dynamic task replication in cyber-physical systems, which resulted in a publication in IEEE Transactions on Emerging Topics in Computing.

Feel free to explore my publications and CV for more details.

selected publications

  1. TETC
    Dynamic Task Replication with Imperfect Fault Detection in Multicore Cyber-Physical Systems
    Hossein Hosseini, Mohsen Ansari, and Jörg Henkel
    IEEE Transactions on Emerging Topics in Computing, 2025
  2. ACL
    IntroLM: Introspective Language Models via Prefilling-Time Self-Evaluation
    Hossein Hosseini, Gholamreza Haffari, Chris Leckie, and 1 more author
    64th Annual Meeting of the Association for Computational Linguistics (Under Review), 2026