Andrei Sirazitdinov
ABOUT ME
Ph.D. in Computer Science
University of Heidelberg | Data Science, ML, AI, Computer Vision
As a researcher, I was focused on causal inference and individualized treatment effect estimation with deep learning. I also worked on explainable AI for healthcare and vision tasks. Permanent residency and full work authorization in Germany. Open to relocation.
Research Interests
Causal inference, individualized treatment assignment, explainable ML
Built and evaluated deep learning models (MLPs, GANs, VAEs, GNNs) in TensorFlow, Keras, and PyTorch for healthcare decision support, privacy, and explainability.
EXPERIENCE
University of Heidelberg, Germany
Apr 2020 - Oct 2025 | Ph.D. Candidate
- • Specialized in causal inference, developing and evaluating individualized treatment assignment strategies using deep learning algorithms in TensorFlow and Keras, including MLPs, GANs, Variational Autoencoders, and Graph Neural Networks (GNNs) (GitHub).
- • Designed and validated pain patient treatment assignment strategies using K-NN clustering and XGBoost, supported by RCT validation. Developed REDCap-based data collection and R Shiny patient characteristic visualizations (GitHub).
- • Built explainable pathology patch classification with prototype learning and decision trees in PyTorch.
- • Applied digital twin models using stable diffusion for patient data privacy in PyTorch.
- • Developed a dropout prediction model for cancer patients achieving 80% precision in TensorFlow.
National Institute of Informatics, Japan
Oct 2018 - Apr 2019 | Research Intern
- • Developed and evaluated a computer vision algorithm for long-term video prediction using PyTorch.
Irkutsk Branch of MSTUCA, Russia
Oct 2014 - Jul 2015 | Student Assistant
- • Built a multi-camera system to capture helicopter panel data and populate digital tables automatically.
EDUCATION
University of Heidelberg, Germany
2020 - 2025 | Ph.D. in Computer Science
Saarland University, Germany
2016 - 2019 | M.Sc. in Visual Computing
Irkutsk State University, Russia
2012 - 2016 | B.Sc. in Applied Informatics and Mathematics
High School, Irkutsk, Russia
2010 - 2012
PROJECTS
Trading Bots (Python)
WebSocket API | Docker | Git | GitHub Actions
Development and server deployment of trading bots in Python using WebSocket API, Docker, Git, and GitHub Actions.
Real-Time ML Platform
Kafka | Apache Iceberg | BERT | REST + Async Pipelines
Built an end-to-end real-time ML platform with Kafka ingestion, Apache Iceberg medallion architecture (bronze/silver/gold), and BERT-based sentiment inference supporting both REST (low-latency) and Kafka-based asynchronous pipelines; integrated a Streamlit UI for real-time visualization (GitHub).
Transformer Text Generation (PyTorch)
Attention | Residual Blocks | Autoregressive Training
Implemented and trained a transformer-based text generation model from scratch in PyTorch, incorporating attention residual blocks and a TARNet-inspired architecture for representation-output separation; designed an autoregressive training pipeline and prepared integration into a real-time ML system (GitHub).
SKILLS
ML/AI and Data
Research and applied machine learning
End-to-end model development, evaluation, and explainability across healthcare and vision tasks.
- TensorFlow
- PyTorch
- Scikit-learn
- Keras
- Pandas
- NumPy
Tools and Platforms
- Python
- SQL
- R
- Docker
- Kubernetes
- MLflow
- Azure (beginner)
- Git
- Linux
- Windows
LANGUAGES
- English (Fluent speaking, reading, and writing)
- German (B1 certificate, completed C1 courses)
- Russian (Native)
SELECTED PUBLICATIONS
- A. Sirazitdinov et al., "Graph Neural Networks for Individual Treatment Effect Estimation," IEEE Access, 2024.
- A. Sirazitdinov et al., "Review of Deep Learning Methods for Individual Treatment Effect Estimation with Automatic Hyperparameter Optimization," TechRxiv, 2022.
CONTACT
- Mannheim, Germany
- dyh@list.ru
- +49 176 47699707
CHAT
Ask the CV Assistant
Note: the chat requires a backend service to respond.