Senior Research Scientist with 14+ years of experience developing and deploying software for complex, data-intensive problems in experimental physics. My work combines scientific computing, applied deep learning, high-performance computing, research leadership, and the delivery of production scientific software.
Professional experience
Senior Research Scientist, AI/ML NICPB/KBFI 07/2020 – Present
High-Energy Physics and Computation Group Tallinn, Estonia
- ML Model Development for Data Reconstruction: Developed novel GNN/Transformer models and data pipelines using PyTorch/TensorFlow on multi-GPU HPC systems (AMD and Nvidia). Deployed with ONNX to production-ready systems. Improved critical accuracy metrics by up to 30%. Led work on multiple peer-reviewed publications, including Nature Communications Physics.
- Software & Performance Optimization: Co-led the reconstruction software group for the CERN CMS collaboration. Optimized the CMSSW C++ codebase and data processing workflows, reducing CPU time for data-taking by 40% in key workflows. Developed the use CI/CD, modern debugging and regression analysis best practices.
- Project Lead & PI - Applied ML Research: Led applied ML research initiatives across different domains, primarily perception in point cloud & image data. Developed ML and data architecture & code, directed research strategy, managed project timelines, reported on deliverables. Focus on CMS and FCC applications.
- Mentorship & Team Leadership: Mentored and supervised three PhD researchers, one MSc researcher, and two BSc researchers. Provided technical guidance and support for ML, simulation, and data engineering tasks contributing to >8 joint peer-reviewed publications.
- Computing Operations: Managed the planning, funding acquisition, procurement, and deployment for a renewal of on-premises HPC infrastructure. DevOps and user support responsibilities for 24/7 operations.
Postdoctoral Researcher Caltech 07/2018 – 06/2020
Experimental High-Energy Physics Group Pasadena, CA, USA
- Data Analysis Pipeline Optimization: Re-engineered data analysis pipelines using CUDA, Python and C++, accelerating time-to-insight by 10x for large-scale columnar datasets.
- HPC DevOps & Reliability: Led DevOps for Caltech’s HPC center supporting critical CERN workloads. Ensured 24/7 system reliability and efficient operation of the batch queues and distributed storage (Ceph/Hadoop).
PhD Researcher ETH Zürich 09/2014 – 06/2018
Experimental High-Energy Physics Group Zürich, Switzerland
- ML for Particle Identification: Implemented and deployed improved ML methods using xgboost for particle identification within the CERN production software environment.
- Data Analysis & Discovery Contribution: Developed data analysis software (C++/Python/numpy) for CERN, contributing to the first observation of the ttH process and heavy-flavour jet identification. Managed research goals and deliverables.
Internship Lingvist Technologies 05/2017 – 07/2017
Data Science Team Tallinn, Estonia
- Predictive Modeling: Developed an LSTM-based model based on business requirements that significantly improved language learning recall analysis in open-ended vocabulary data streams.
Research Engineer NICPB/KBFI 01/2012 – 08/2014
High-Energy Physics and Computation Group Tallinn, Estonia
- CERN Data Analysis: Developed data analysis software (C++/Python/numpy) software development for the CMS experiment at CERN.
Education
- PhD, Experimental Particle Physics (ETH Medal), ETH Zürich (Thesis) 09/2014 – 07/2018
- M.Sc., Fundamental Physics (cum laude), University of Tartu, Estonia 09/2012 – 06/2014
- B.Sc., Physics, University of Tartu, Estonia 09/2008 – 06/2012
Technical skills
- Software Engineering: CI/CD (Github/Gitlab), debugging, optimization, regression analysis 14 years
- HPC: distributed storage (CEPH, Hadoop) and processing (Slurm), NoSQL datasets (ROOT, parquet), DevOps (ansible) 12 years
- Software Development: Python, C/C++ in production, legacy code maintenance and modernization 10 years
- ML: pytorch & tensorflow deployment, use of multi-GPU systems including HPC 7 years
Core competencies
- Scientific Computing & Software Engineering: familiarity with the CERN stack (C++, Python, libraries & distribution) 14 years
- Data Analysis and Engineering: contribution to multiple key measurements at CERN through large-scale data analysis 11 years
- Physics R&D: 10+ peer-reviewed research results 11 years
- Applied ML/AI R&D: 8+ peer-reviewed papers on applied AI methods 7 years
- Project Management: research funding acquisition/reporting, computing software and infrastructure 6 years
- Technical Leadership: successfully led PhD and MSci projects, co-led the CERN CMS reconstruction team 5 years
Industry projects
- Mu-Ray Tech: advisor 2025 – Present
- Taara Robotics: real-time multi-task segmentation and object identification networks for Jetson Orin NX DLA 2025
- GScan: accurate tomography reconstruction using 3D-CNNs for construction safety 2023
Language skills
- English full proficiency
- Russian, French limited working
- Korean elementary
- Estonian native language
Scientific publications and reports
- Seeba, N.-N. et al, “ParticleTransformer is all you need for reconstructing hadronic tau leptons”, arXiv:2606.18460 (2026)
- Mokhtar, F. et al, “Machine-learned particle flow as a foundation model for collider physics”, arXiv:2606.14373 (2026)
- CMS collaboration, “Full event interpretation with machine-learning-based particle-flow reconstruction in the CMS detector”, Eur. Phys. J. C (2026), 10.48550/arXiv.2601.17554
- Tani, L. et al, “Reconstructing hadronically decaying tau leptons with a jet foundation model”, SciPost Physics (2025), 10.21468/SciPostPhysCore.8.3.046
- Mokhtar, F. et al, “Fine-tuning machine-learned particle-flow reconstruction for new detector geometries in future colliders”, PRD (2025), 10.1103/PhysRevD.111.092015
- Põder, S. et al, “On the detection of stellar wakes in the Milky Way: a deep learning approach”, Astronomy and Astrophysics (2025), 10.1051/0004-6361/202451480
- Tani, L. et al, “A unified machine learning approach for reconstructing hadronically decaying tau leptons”, Computer Physics Communications 307 (2025), 10.1016/j.cpc.2024.109399
- Pata, J. et al, “Improved particle-flow event reconstruction with scalable neural networks for current and future particle detectors”, Nature Communications Physics 7 (2024), 10.1038/s42005-024-01599-5
- Lange, T. et al, “Tau lepton identification and reconstruction: a new frontier for jet-tagging ML algorithms”, Computer Physics Communications 298 (2024), 10.1016/j.cpc.2024.109095
- CMS Collaboration, “Progress towards an improved particle flow algorithm at CMS with machine learning”, ACAT (2022), CERN-CMS-DP-2022-061
- Lewicki, M. et al, “Dynamics of false vacuum bubbles with trapped particles”, Phys.Rev.D 108 (2023), https://doi.org/10.1103/PhysRevD.108.036023
- Põder, S. et al “A Bayesian estimation of the Milky Way’s circular velocity curve using Gaia DR3”, Astronomy and Astrophysics 676 (2023), 10.1051/0004-6361/202346474
- Wulff, E. et al, “Hyperparameter optimization of data-driven AI models on HPC systems”, J.Phys.Conf.Ser. 2438 (2023), 10.1088/1742-6596/2438/1/012092
- Bazarov, A. et al, “Sensitivity Estimation for Dark Matter Subhalos in Synthetic Gaia DR2 using Deep Learning”, Astronomy and Computing (2022), 10.1016/j.ascom.2022.100667
- Pata, J. for the CMS Collaboration, “Machine Learning for Particle Flow Reconstruction at CMS”, ACAT (2022), 10.1088/1742-6596/2438/1/012100
- Pata, J. et al, “MLPF: Efficient machine-learned particle-flow reconstruction using graph neural networks”, EPJC (2021), 10.1140/epjc/s10052-021-09158-w
- Pata, J. et al, “Data Analysis with GPU-Accelerated Kernels”, Proceedings of Science, ICHEP (2020) https://doi.org/10.22323/1.390.0908
- CMS Collaboration, “Observation of ttH production”, PRL (2018), 10.1103/PhysRevLett.120.231801
- CMS Collaboration, “Identification of heavy-flavour jets with the CMS detector in pp collisions at 13 TeV”, JINST (2018), 10.1088/1748-0221/13/05/P05011
Key research talks
- Panel discussion: “Cutting Through the Hype – Quantum and AI Technology Limits, Lessons, and Next Moves”, sTARTUp Day 2026 (Tartu, Estonia) 2026
- Particle flow reconstruction with a learnable, differentiable, efficient ML model, ZPW2026 (Zurich, Switzerland) 2026
- Invited talk on science and society, TeadusEST 2025 (Tartu, Estonia) 2025
- Invited talk, Taltech AI Retreat (Estonia) 2025
- CERN OpenLab workshop, invited talk on machine learning for data reconstruction 2025
- Invited talk, Estonian Academy of Sciences (Tallinn, Estonia) 2025
- Scalable neural networks for event reconstruction, ACAT (Stony Brook, NY, USA) 2024
- Neural networks and terascale datasets for particle-flow reconstruction, ML4Jets (Hamburg, Germany) 2023
- Overview of machine learning for calorimeter clustering and particle flow, Learning To Discover (Paris) 2022
- Machine learning for data reconstruction at the LHC, LIP seminar (Portugal), invited, virtual 2022
- Graph neural networks, QU Data Science Basics (Hamburg), invited 2021
- Machine learning for particle flow reconstruction at CMS, ACAT (Daejeon, South Korea), virtual 2021
- Measurements of ttH at CMS, Lake Louise (Canada) 2019