cv

Senior Data Scientist

Migros-Genosenschafts Bund, Zürich, Switzerland

• Definition of tasks for the team in the agile work environment
• Review of the team members' code
• Mentoring less experienced team members and sharing knowledge in the internal meetup
• An ETL pipeline for data ingestion and processing
  • local python scripts with multi-process parallelisation using pandas + joblib
  • Airflow with python scripts in pods and scala jobs on Spark
• Store revenue modelling using LightGBM and Lasso regressions in python
• User-facing dashboard application using dash running in k8s using a docker image
• Testing using pytest with 60% coverage over 25k+ LoC + code linting and type checks
• Implementation of CI/CD pipelines in gitlab
• A/B testing of the project success

Customer Care Data Scientist

KNIME, Constance, Germany

• Implementation of model interpretation algorithms: K-LIME
• Customer support for KNIME Server and PoC preparation crafted to user needs

Kaggle competitor

• 9th place out of 7000+ teams (top 1% and gold medal) on a kaggle.com competition
• 3 silver medals for kernels with total on 10000 views

Physics Researcher

Universität Heidelberg

• Assessed classifiers of heavy particles exploiting designed features: DNN and GBT have shown twice better performance than historic methods.
• Mined through events on huge datasets of 100+ PB using high-throughput computing with data visualisation on a local machine.
• Developed a software toolkit for the Data Quality Monitoring for a heterogeneous HW system.
• Led a team of developers.

Research Fellow

DESY, Hamburg, Germany

• Advanced parallelisation of high-precision Monte Carlo predictions taking more than 100.000 CPU hours.
• Boosted the speed of production by 60%. Reduced the time of calibration of a key detector by a factor more than 2.
• Led a physics group of 10+ people.

Research Assistant

DESY, Hamburg, Germany

• Engineered a new feature, that allowed to improve physics measurement by factor 2 in precision.
• Leading efforts in Monte Carlo production with the high-throughput computing on the GRID. Doubled the annual event production rate.

PhD

Universität Hamburg, Hamburg, Germany

• Measurement of charm production in deep inelastic scattering using lifetime tagging for D^± meson decays with the ZEUS detector at HERA

MSc

Taras Shevchenko National University of Kyiv