General Information

Full Name Balázs Hidasi
Short summary Research scientist working on the field of machine learning with 15+ years of experience in designing algorithms. Spent most of his career in the intersection of science and industry, researching recommender systems and algorithms.

Scientific profile

  • Overview of main research topics
    • Counterfactual evaluation (2023 - ongoing)
    • Evaluation of recommender systems (2022 - ongoing)
    • Deep learning for recommender systems (2015 – ongoing)
    • Context-aware factorization methods on implicit feedback data (2011 – 2015)
    • Time series classification (2008 – 2011)
  • Participation in the scientific community
    • Regular presenter at scientific conferences and meetups.
    • Peer reviewer for scientific conferences (e.g. RecSys, KDD, UMAP, WSDM, etc.) and journals.
    • Main organizer of the Deep Learning for Recommender Systems (DLRS) workshop series (2016-2018).
    • Co-organizer and recurring presenter of the Budapest Recommender Systems Meetup (2016-2017).
  • Tutorials, talks, teaching
    • Mentoring / knowledge dissemination in the company.
    • Lecture on recommender systems at KÜRT Academy (2018 - 2022, one presentation semiannually)
    • RecSys Summer School 2017 - Deep Learning for Recommender Systems (August 2017)
    • Co-organizer of the tutorial on deep learning for recommender systems at RecSys2017 (August 2017)
    • Context-aware recommendations at the summer school of the University of Szeged (24 July 2014)
    • Research presentation to fellow researchers at the Technical University of Delft (11 April 2014)
    • Lectures on recommender systems at the Budapest University of Technology and Economics (2011-2016)


  • Research: deep learning | recommender systems | recurrent neural networks | tensor and matrix factorization | collaborative filtering | implicit feedback | context-awareness | counterfactual learning | reinforcement learning | algorithm design | machine learning | generative AI
  • Programing / technology: python | scipy stack (numpy, pandas, scipy, sklearn, etc.) | Theano | PyTorch | Tensorflow | basics of JAX | Java | C++ | CUDA | SQL | git
  • Languages: Hungarian (native) | English (full professional proficiency) | German (elementary proficiency)


  • 2022.07 -
    Gravity R&D, a Taboola Company
    • After the acquisition by Taboola, my main objective is to figure out (1) the best way of co-operation between my team of 3 machine learning researchers and engineers and the company's algorithms department and other teams; (2) the best way to make impact in this new setting.
    • Highlights of my team's contributions include
      • (a) Getting back to publishing scientific work after a few years of hiatus.
      • (b) By extending our production deep learning framework and improving Gravity's CTR/CVR prediction algorithms, we achieved ~10%+10% improvement in key metrics.
      • (c) Designing and productionizing an innovative solution for e-commerce creative generation based on generative AI technology that increases key metrics by ~15%.
      • (d) Providing machine learning expertise whereever it is needed.
  • 2015.01 - 2022.07
    Head of Research
    Gravity Research and Development
    • My role as Head of Research was to oversee the research efforts of the company, as well as to conduct my own research. I was also responsible for providing machine learning expertise to any of the ongoing projects of the company.
    • Highlights of my work are
      • (a) Creating the GRU4Rec algorithm (family) that improved upon our previous solution by 10-20% in the revenue through recommendations.
      • (b) Laying down the basics of our CTR/CVR prediction framework that contributed to the success of our co-operation with Taboola that eventually led to the acquisition of Gravity R&D by Taboola.
      • (c) Maintaining and increasing the company's renown in the scientific community via high quality published research, event organization, tutorials, and presenting at meetups and other invited talks.
      • (d) Optimizing off-the-shelf deep learning frameworks with custom CUDA operators achieving up to 100x speed up for certain operations.
  • 2021.01 - 2022.07
    Leader of the Deep Learning Team
    Gravity Research and Development
    • My role was to build up and lead a new team focusing on improving our deep learning based recommender solutions.
    • I built and managed a team of 2 machine learning engineers.
      • My team completely overhauled our production deep learning training and inference framework, improving its flexibility, increasing its efficiency, and extending its feature set.
  • 2015.01 - 2020.01
    Leader of the Data Mining Team
    Gravity Research and Development
    • My role, as the leader of the data science team, was to oversee all data science related tasks in the company.
    • I built and managed a team of 3 data scientists. My team
      • (a) applied my research results in the production system
      • (b) significantly improved data infrastructure of the company
      • (c) performed fine tuning during POCs, so that the company has never lost a single A/B test against its competitors
      • Beside these achievements, we upheld a healthy work-life balance in the team and that all members of my team enjoyed their time spent there.
  • 2010.01 - 2015.01
    Data Mining Researcher
    Gravity Research and Development
    • My main responsibilities, as a data mining researcher, were researching new recommender algorithms, putting algorithms into production, analyzing user behavior data, and fine tuning our recommendation logic.
    • Researched context-aware factorization methods for implicit feedback data.
    • Core member of the EU FP7 funded CrowdRec research project (2013-2016).
  • 2015.06 - 2015.09
    Visiting Researcher
    Telefónica I+D
    • Research collaboration between Gravity R&D & Telefónica I+D in the CrowdRec project.
    • Laid the foundations of GRU4Rec.
  • 2008.01 - 2011.09
    Individual Researcher
    DmLab, TMIT, BME-VIK
    • Working on the ShiftTree algorithm.
    • I was affiliated with the university research group, DmLab.


  • 2014.09 - 2016.06
    PhD candidate
    Budapest University of Technology and Economics, Budapest, Hungary
    • Data Science and Content Technologies Laboratory (DCLab)
    • Summa cum laude Ph.D. (30 June 2016)
  • 2011.09 - 2014.09
    PhD studies
    Budapest University of Technology and Economics, Budapest, Hungary
    • Computer Sciences Doctorate School
    • Intelligent Systems Group
  • 2009.02 - 2011.07
    Budapest University of Technology and Economics, Budapest, Hungary
    • Faculty of Electrical Engineering and Informatics
    • Conputer Science and Engineering
    • Graduated with highest honors (21 June 2011)
  • 2005.09 - 2009.02
    Budapest University of Technology and Economics, Budapest, Hungary
    • Faculty of Electrical Engineering and Informatics
    • Conputer Science and Engineering
    • Graduated with highest honors (08 January 2009)