Oğuz Güzel

Experimental particle physics leveraging machine learning with data collected at CERN.

About me

PhD candidate in experimental particle physics who applies engineering discipline to machine learning, scalable Python software, and data-driven decision workflows. I translate complex data and model validation into reliable, reproducible systems for research and operational use.

Summary

I bring research-level rigor to industry-style ML and data engineering. My work covers end-to-end workflow design, model benchmarking, automated testing, and production-ready tooling for large-scale datasets.

Applied ML and software delivery

  • Designed and benchmarked PyTorch models for complex, high-dimensional datasets, including transformers, deep neural networks, and decision-tree ensembles.
  • Developed representation-learning prototypes using variational autoencoders (VAEs) to improve feature extraction and model robustness.
  • Built reproducible pipelines with Docker, CI/CD, automated validation, and data orchestration for scalable experimentation.
  • Produced clear performance metrics, evaluation reports, and documentation for stakeholder decision making.
  • Collaborated with distributed teams on large scientific software projects, applying Agile-style communication and version-controlled delivery.

Projects and portfolio

I maintain a portfolio of data science and ML work, including models that combine transformer architectures with physics-driven feature engineering.

  • Transformer Classifier: a PyTorch transformer encoder model for signal/background classification in collider data. GitHub
  • XGBoost Classifier: a notebook-driven XGBoost workflow for HH → bbWW classification using particle physics features. GitHub
  • Check my github repositories for more: https://github.com/Oguz-Guzel

Talks and presentations

Selected presentations that bridge technical model work and data-driven physics analysis:

  • “EFT interpretations in the Higgs sector at the CMS experiment” — DIS2024, Grenoble, FRANCE. Details
  • “Higgs self-coupling measurements at the CMS experiment” — PASCOS 2024, Quy Nhon, VIETNAM. Program

Teaching and mentoring

  • Physics lab tutor, electromagnetism course at Istanbul Technical University. Supported undergraduate lab training, experiment setup, and data analysis practice.

Publications

I maintain a curated publications page for selected CMS contributions and point to the broader collaboration list for full context.

Education

  • PhD in Physics, UCLouvain — 2022 to expected Sept 2026
  • MSc in Physics, Istanbul Technical University — 2019 to 2022
  • BSc in Astronautical Engineering, Istanbul Technical University — 2014 to 2019

Skills

  • Python, PyTorch, NumPy, pandas, transformer architectures
  • Machine learning validation, robustness, and production-readiness
  • Docker, CI/CD, scalable data pipelines, reproducible tooling
  • Clear technical communication, cross-team collaboration, mentoring

Online