About Me

I am a final-year PhD candidate in the School of Informatics at the University of Edinburgh, advised by Dr. Marc Juarez. My work focuses on trustworthy AI/ML, in particular privacy, security and fairness of machine learning including large language models (LLMs). I started my PhD at RISE Research Institutes of Sweden and later moved to the University of Edinburgh, where I will complete my degree.

During my PhD, my research has focused on reconciling privacy with fairness and accountability in distributed ML under real-world constraints . More recently, I started working on the trustworthiness of LLM-based systems, especially their fairness and security, in high-stakes domains such as hiring pipelines, in collaboration with Amazon.

Projects & Publications

Ongoing

Trustworthiness in LLM-based systems for high-stakes domains

School of Informatics, The University of Edinburgh · Edinburgh, United Kingdom

LLM-based systems are increasingly deployed in high-stakes domains such as hiring pipelines, where their decisions directly affect people. This project studies how trustworthy these systems are in such settings with a focus on their fairness and security. It is funded by and carried out in collaboration with Amazon.

Bias and fairness in large language models (LLMs)

School of Informatics, The University of Edinburgh · Edinburgh, United Kingdom

Bias in an LLM is not spread evenly across the network but concentrated in particular components. This work first locates which attention heads and feed-forward layers carry it, then fine-tunes only those components against a fairness-aware training objective.

Completed

Privacy and fairness tensions in distributed ML

  1. Under review

    Group Fairness in Differentially Private Federated Learning

    Federated learning exhibits performance disparities across demographic groups due to data heterogeneity over cleints, and differential privacy can compound them. To address this, we designed a group-aware model architecture that reduces performance gaps across groups under differential privacy without sharing group information.

  2. IJCNN '25

    Empirical Analysis of Asynchronous Federated Learning on Heterogeneous Devices: Efficiency, Fairness, and Privacy Trade-offs

    S. Mohammadi, et al.

    IJCNN '25: 2025 International Joint Conference on Neural Networks

    PDF arXiv Code

Scalable and efficient privacy preservation in distributed ML

  1. PMCJ '25

    EncCluster: Scalable Functional Encryption through Weight Clustering and Probabilistic Filters

    V. Tsouvalas*, S. Mohammadi*, et al.

    PMCJ '25: Pervasive and Mobile Computing (Elsevier)

    PDF arXiv Code
  2. NeurIPS W '24

    Bringing Functional Encryption to Federated Foundation Models

    V. Tsouvalas*, S. Mohammadi*, et al.

    Workshop at NeurIPS '24: International Workshop on Federated Foundation Models in Conjunction with NeurIPS 2024 (FL@FM-NeurIPS'24)

    PDF arXiv Code
  3. FedCSIS '23

    Balancing Privacy and Accuracy in Federated Learning for Speech Emotion Recognition

    S. Mohammadi, et al.

    FedCSIS '23: Proceedings of the 18th Conference on Computer Science and Intelligence Systems

    🏆 Professor Zdzisław Pawlak Award
    PDF Code

Private machine unlearning in distributed settings

  1. CIKM '25

    EFU: Enforcing Federated Unlearning via Functional Encryption

    S. Mohammadi*, V. Tsouvalas*, et al.

    CIKM '25: Proceedings of the 34th ACM International Conference on Information and Knowledge Management

    PDF arXiv Code

* Equal contribution. Full publication list on Google Scholar →

Reach me here