Drazen Zaric, Data Scientist and Machine Learning Developer in Belgrade, Serbia
Drazen Zaric

Data Scientist and Machine Learning Developer in Belgrade, Serbia

Member since May 30, 2019
Drazen is a data scientist and data engineer with over seven years of experience building analytics infrastructure, machine learning, and data analysis. He has extensive experience in big data, machine learning, and web development. He has worked with notable companies such as Slack and Microsoft, and built products used by millions of users.
Drazen is now available for hire


  • Freelance Consultant
    SQL, Big Data, Tableau
  • Nordeus
    Java, Python, JavaScript, Aerospike, Hadoop, Apache Impala
  • Nordeus
    Python, Machine Learning, SQL, Tableau, Flask, Scikit-learn, Numpy, Pandas



Belgrade, Serbia



Preferred Environment

MacOS, Linux, Python, Vim

The most amazing...

...project I've built was an analytics infrastructure for a product with 150 million users.


  • Analytics Consultant

    2019 - 2019
    Freelance Consultant
    • Collaborated with small companies on establishing their data infrastructure, discipline, and teams.
    • Assessed existing processes, available data, and infrastructure.
    • Planned for future growth in terms of product, and the team.
    • Helped with descriptive and predictive user behavior models, segmentation, and A/B testing.
    Technologies: SQL, Big Data, Tableau
  • Product Manager

    2017 - 2018
    • Led the team in developing Golden Boot, a game for a then new platform called Facebook Instant Games. This product grew from an idea to over 10 million players in four months.
    • Oversaw the product and project management, back-end development, data analysis, and game design within an Agile team.
    • Developed high performance algorithms for matchmaking.
    • Developed a distributed push notifications system backed by a Hadoop data warehouse and a machine learning model for user churn.
    Technologies: Java, Python, JavaScript, Aerospike, Hadoop, Apache Impala
  • Data Scientist

    2015 - 2017
    • Developed internal data products. Designed, developed, and deployed optimization algorithms for in-game tournament generation.
    • Developed machine learning models for anomaly detection, product performance and user behavior prediction, tools for optimizing marketing campaigns.
    • Maintained and improved a custom, in-house BI tool.
    • Supported product managers, game designers, and user acquisition teams with analyses, and insights.
    Technologies: Python, Machine Learning, SQL, Tableau, Flask, Scikit-learn, Numpy, Pandas
  • Data Engineer

    2013 - 2015
    • Developed and maintained an in-house ETL system, and data warehouse on top of Hadoop and Impala for TopEleven, an online game with over 150 million players.
    • Transitioned a single PostgreSQL instance to an in-house Hadoop cluster solution.
    • Designed and wrote data import and SQL transformations, and oversaw the data quality.
    • Integrated data from internal and third-party data sources.
    • Scaled data warehouse and processes from a single product, to multiple products.
    Technologies: Python, SQL, Hadoop
  • Software Development Engineer

    2011 - 2013
    • Developed the PDF Reflow feature for Word 2013.
    • Designed algorithms for converting a PDF into an editable DOCX document.
    • Implemented and evaluated machine learning models for identifying parts of a PDF document.
    Technologies: C++


  • Top Eleven by Nordeus (Development)

    A highly successful online sports games, Top Eleven has had over 150 million players to date. I worked on the analytics infrastructure, product management, data science, and in-game algorithm development.

  • Votum Polls for Slack (Development)

    This is a Slack app I developed end-to-end, enabling the polling and running of idea tournaments within Slack.

  • Selecta: An Amarok Player Extension (Development)

    This is an inverted index and tf-idf retrieval engine for music in your library based off Last.fm tags. This was implemented from scratch in JavaScript, done back in the days when music was offline.

  • Improved Heatmaps in Python (Development)

    This is a Python code, and a Jupyter notebook for an improved heatmap implementation using Matplotlib and Seaborn.

    Similar to what you can easily get in Tableau using a Size parameter, here you can have a square as a parameter depending on the field value.

    Described in detail in this blog post: https://towardsdatascience.com/better-heatmaps-and-correlation-matrix-plots-in-python-41445d0f2bec


  • Languages

    Python, SQL, Java
  • Frameworks

    Flask, Hadoop
  • Libraries/APIs

    Scikit-learn, Pandas
  • Tools

    Apache Impala, Tableau, Ansible
  • Paradigms

    Data Science
  • Other

    Machine Learning, Algorithms, Data Visualization, Optimization
  • Platforms

    Linux, Apache Kafka
  • Storage

  • Industry Expertise

    Web Development


  • Master's degree of Science in Computer Science
    2010 - 2011
    University of Novi Sad - Novi Sad, Serbia
  • Bachelor's degree in Computer Science
    2006 - 2010
    University of Novi Sad - Novi Sad, Serbia


  • Product Management Program
    UC Berkeley Executive Education

To view more profiles

Join Toptal
Share it with others