CV

General Information

Full Name Amir Joudaki
Title Fundamental AI/ML researcher
Location Zurich, Switzerland
LinkedIn https://linkedin.com/in/AmirJoudaki/
GitHub https://github.com/ajoudaki/
Google Scholar https://scholar.google.com/citations?user=iCXSFxMAAAAJ

Practical Experience

  • Nov 2024 – present
    Postoctoral researcher
    ETH Zurich
    • Working towards a comprehensive understanding of how deep nonlinear models learn
    • Researching barriers to learning in real world tasks with task and distributional changes, as well as reasoning
    • Supervising multiple MSc students
  • Nov 2024 – present
    Lead Research Collaborator
    ETH Zurich & Apple Inc.
    • Contributing to fundamental research on barriers for learning in real world applications where distributions shift
    • Developed a novel mathematical framework to predict how and why deep neural networks lose their ability to adapt
    • Design and implementation of comprehensive experiments on wide-ranging tasks and architectures
  • Jan 2022 – July 2022
    Machine Learning Consultant
    SkillLab
    • Research and development of a proof-of-concept for training, labeling, and monitoring NLP tasks
    • Written in PyTorch, Flask, Typescript, React, MongoDB, orchestrated as micro-services in REST API
    • Run inside Docker containers for deployment

Education

  • Sept 2019 – Nov 2024
    Ph.D. in Artificial Intelligence
    ETH Zurich
    • Authored 8 papers (6 first-authored) in top-tier venues in AI/ML and computational genomics
    • Supervised >10 MSc thesis/semester projects, leading to several top-tier publications
    • Organized teaching
  • Feb 2017 – Nov 2024
    Direct Ph.D. (MSc + Ph.D.) in Artificial Intelligence
    ETH Zurich
    • Direct PhD is a highly selective program (<5% of all PhD admissions)
    • 2 years of MSc courses followed by 4 years of PhD research
    • Ph.D. supervisors: Gunnar Ratsch & Francis Bach
    • MSc Thesis: Scalable algorithms for biological sequence analysis
  • Feb 2014 – Jan 2017
    M.Sc. in Cognitive Neuroscience
    International School for Advanced Studies (SISSA)
    • Thesis: Modeling activity of electrophysiological recordings in vivo in rats
  • Sept 2008 – Sept 2011
    B.Sc. in Computer Engineering
    Sharif University of Technology

Programming Skills

  • Python · PyTorch · JAX · Scikit-Learn · NumPy
  • C/C++ · OpenMP · MPI · CUDA · Numba
  • Docker · Spark · SQL · Bash · Git · Java · LaTeX

Publications

  • Mathematical foundations of AI
    • Amir Joudaki, Giulia Lanzillotta, Mohammad Samragh Razlighi, Iman Mirzadeh, Keivan Alizadeh, Thomas Hofmann, Mehrdad Farajtabar, Fartash Faghri. 'Barriers for Learning in an Evolving World: a Mathematical Understanding of Loss of Plasticity' (submitted, under review)
    • Amir Joudaki, Thomas Hofmann. 'Emergence of globally attracting fixed points in deep neural networks with nonlinear activations', AISTATS (poster)
    • Alex Meterez*, Amir Joudaki*, Francesco Orabona, Alex Immer, Gunnar Ratsch, Hadi Daneshmand. 'Batch normalization without gradient explosion: Towards training without depth limits', ICLR 2024 (poster). (*equally contributed)
    • Amir Joudaki, Hadi Daneshmand, Francis Bach. 'On the impact of activation and normalization in obtaining isometric embeddings at initialization', NeurIPS 2023 (poster)
    • Amir Joudaki, Hadi Daneshmand, Francis Bach. 'On Bridging the Gap between Mean Field and Finite Width in Deep Random Neural Networks with Batch Normalization', ICML 2023 (poster)
    • Hadi Daneshmand, Amir Joudaki, Francis Bach. 'Batch Normalization Orthogonalizes Representations in Deep Random Networks', spotlighted NeurIPS 2021 (top 3% of submissions)
    • Alexandre Bense, Amir Joudaki, Tim G. J. Rudner, Vincent Fortuin. 'PCA Subspaces Are Not Always Optimal for Bayesian Learning', NeurIPS 2021 workshop (DistShift)
  • Genomics Sequence Analysis
    • Flavia Pedrocchi, Stefan Stark, Gunnar Ratsch, Amir Joudaki. 'Identifying Biological Priors and Structure in Single-Cell Foundation Models', ICML workshop 2024 (Efficient and Accessible Foundation Models for Biological Discovery)
    • Kacper Kapusniak, Manuel Burger, Gunnar Ratsch, Amir Joudaki. 'Learning Genomic Sequence Representations using Graph Neural Networks over De Bruijn Graphs', NeurIPS workshop 2023 (Graph Learning Frontiers)
    • Amir Joudaki*, Alexandru Meterez*, Harun Mustafa, Ragnar Groot Koerkamp, Andre Kahles, Gunnar Raetsch. 'Aligning distant sequences to graphs using long seed sketches', Genome Research (Cold Spring Harbor Laboratory Press), 2023. (*equally contributed)
    • Amir Joudaki, Gunnar Ratsch, Andre Kahles. 'Fast Alignment-Free Similarity Estimation By Tensor Sketching', proceedings of RECOMB 2021
    • Mikhail Karasikov, Harun Mustafa, Amir Joudaki, Sara Javadzadeh-No, Gunnar Rätsch, André Kahles. 'Sparse Binary Relation Representations for Genome Graph', Journal of Computational Biology 27(4) (2020): 626–639
  • Dimensionality Reduction
    • Amir Najafi, Amir Joudaki. 'Nonlinear Dimensionality Reduction via Path-Based Isometric Mapping', IEEE TPAMI 38(7): 1452–1464
  • Functional Brain Networks Analysis
    • Amir Joudaki*, Elham Barzegaran*, Mahdi Jalili*, Andrea O. Rossetti, Richard S. Frackowiak, Maria G. Knyazeva. 'Properties of functional brain networks affect frequency of psychogenic non-epileptic seizures', Frontiers in Human Neuroscience 6:335. (*equally contributed)
    • Amir Joudaki, Niloufar Salehi, Mahdi Jalili, Maria Knyazeva. 'EEG-based functional brain networks: does the network size matter?', PLOS ONE 7(4): e35673

Supervision / Mentorship

  • Current students
    • Guiv Farmanfarmayan — 'Improving the efficiency of RL-based reasoning'
    • Annalisa Belloni — 'Improving cloning models with insights from loss of plasticity'
    • Flavia Pedrocchi — 'Interpreting single cell foundation models using sparse auto-encoders'
  • Former students and mentorships
    • Alex Meterz — PhD student at Harvard University
    • Alec Flowers — Nvidia
    • Kacper Kapusniak — PhD student at Oxford University

Awards

  • 2025
    • Best Reviewer Award — Best reviewer AISTATS 2025
    • Top reviewer ICML 2025
  • 2017
    • Direct Doctorate Fellowship — Among two out of more than one hundred students selected for direct PhD program at ETH Computer Science Department
  • 2011
    • Ranked 42nd — Among more than 50,000 participants in National Higher Education Entrance Exam
  • 2007
    • Ranked 369th — Among more than 400,000 participants in National University Entrance Exam