Al Depope

Curriculum Vitae

Postdoctoral Researcher in statistical genetics and machine learning at ISTA. Croatian EU citizen.

Download PDF al.depope@ist.ac.at ORCID

Profile

I develop mathematically grounded machine learning for genomics: Bayesian models and inference procedures that improve biomarker selection and individual risk prediction, combined with the high-performance computing (C++, CUDA, distributed systems) and data analysis (Python, R) needed to run them at biobank scale. I am looking for a computational biology position where this work turns into tools with real-world impact.

Experience

Jul 2026 – present

Postdoctoral Researcher

Institute of Science and Technology Austria (ISTA) · Klosterneuburg, Austria

Extending my doctoral work by developing TLgVAMP, a transfer-learning framework for improved cross-ancestry polygenic risk scores.

Sep 2020 – Jul 2026

PhD Candidate

Institute of Science and Technology Austria (ISTA) · Klosterneuburg, Austria

Engineered high-performance approximate message passing frameworks for Bayesian inference in C++ and Python, achieving 10× speed-ups over traditional MCMC methods while maintaining superior variable selection accuracy.

  • Scalable high-dimensional genomic inference. Developed and open-sourced gVAMP and gVAMPomi. Processed 17 million single nucleotide polymorphisms to execute the largest joint GWAS to date, achieving a state-of-the-art 46% prediction accuracy for human height.
  • Proteomic survival analysis. Built vampW, a scalable Bayesian framework modelling disease onset times. Applied to the UK Biobank Pharma Proteomics dataset, achieving a 26–33% relative improvement in onset prediction over penalised Cox and baseline deep-learning approaches.
  • Cross-field collaboration. Partnered with the Textile Recycling Group at TU Wien on targeted statistical analyses and data visualisations supporting their engineering processes.

Relevant coursework: pharmacoinformatics, probabilistic graphical models, deep-learning topics in computer-aided drug design.

Nov 2024 – Nov 2025

Machine Learning & Privacy Intern

University of Vienna · Vienna, Austria

  • Developed a modular Python library for benchmarking membership inference attacks (MIA) on large language models.
  • Designed and implemented a robust document-level differential privacy auditing framework.
Jul 2019 – Aug 2019

Software Engineer Intern

CERN (European Organization for Nuclear Research) · Geneva, Switzerland

  • Integrated 11 environmental data pipelines into the CERN CMS Online Monitoring System.
  • Developed 13 Java aggregation-layer endpoints and 16 Python presentation probes using Plotly.js, optimising latency and automating detector on-call reporting.

Technical skills

Programming
Python (pandas, PyTorch, Hugging Face, scikit-learn), C++ (OpenMP, MPI, Eigen), R, CUDA, SQL, Bash
Data science & ML
High-dimensional Bayesian inference, statistics, probabilistic predictive modelling, numerical optimisation, survival analysis
Bioinformatics
Large-scale GWAS, single-cell foundation models, Nextflow, DNAnexus
DevOps & tools
Git, Docker, SLURM, high-performance computing clusters, Claude Code

Education

Sep 2020 – Jul 2026

PhD in Statistical Genetics

Institute of Science and Technology Austria (ISTA)

Sep 2018 – Sep 2020

MSc in Mathematical Statistics

University of Zagreb · GPA 5.0 / 5.0

Sep 2015 – Sep 2018

BSc in Mathematics

University of Zagreb · GPA 5.0 / 5.0

Transferable skills

Key publications

The full list, with preprints, code and talks, is on the publications page.

Languages & interests

Languages
Croatian (native) · English (fluent) · German (A2)
Work eligibility
Croatian EU citizen
Interests
Hiking — most recent peak above 4,000 m: Mauna Kea, Hawaii — running, and volunteering as a lecturer in competitive mathematics for gifted high-school students.

References