Katia Hayati, Mathematics Developer in Sunnyvale, CA, United States
Katia Hayati

Mathematics Developer in Sunnyvale, CA, United States

Member since October 24, 2016
Katia has worked for twelve years as an engineer, hands-on engineering manager, and consultant. She has worked at companies of all sizes, from tiny two-person startups to major enterprises including Groupon and Microsoft. She specializes in data processing, data-related algorithms, searching, and machine learning, but she has worked up and down the whole stack.
Katia is now available for hire



  • Mathematics 20 years
  • Statistics 20 years
  • Perl 18 years
  • Data Processing 14 years
  • SQL 12 years
  • A/B Testing 10 years
  • Analytics 6 years
  • Data Science 6 years


Sunnyvale, CA, United States



Preferred Environment

*nix, Git

The most amazing...

...feedback I've ever received is a user writing in to thank us for saving his life.


  • Data Science and Machine Learning Consultant

    2013 - PRESENT
    • Worked independently or with an established team, writing production code or mining offline data, I help companies crack their tough data-related problems.
    • Defined and implemented a quality evaluation process for entity matching for Radius Intelligence. Designed, implemented, and launched an algorithm improvement to halve recall errors while keeping precision constant. Worked with the VP of Engineering and CTO to define and interview for a data science engineer position.
    • Helped validate feasibility of a new line of business for Agari by building and evaluating a machine learning model. Bootstrapped a supervised classifier with unsupervised learning to overcome the absence of labeled data and data labeling resources.
    • Defined and prototyped a provably fair advertising revenue attribution model for Sojern. Worked with the CTO to design an A/B testing framework.
    • Improved a credit card fraud detection algorithm, saving Groupon over a million dollars annually by reducing false positives while keeping detection rates high.
    Technologies: Java, Python, Ruby, Perl, R
  • Senior Engineering Manager

    2011 - 2012
    • Worked with sales, city planning, engineering and data science to build a demand forecasting system.
    • Built an email A/B testing framework and reduced outages due to setup errors from several a day to 0.
    • Significantly increased revenue and user engagement by continuously delivering relevance and UI improvements.
    Technologies: Java, Perl, Ruby, Ruby on Rails
  • VP of Engineering

    2010 - 2011
    • Reported directly to the CEO as the first employee hired at a social video marketing startup.
    • Hired and managed a team of back-end and front-end engineers.
    • Helped grow the client list from two to over sixty and page views from a handful to several million a day.
    • Worked with the Chief Architect to design and implement infrastructure, including fully customizable video recording, internationalization, analytics processing, video search, and smart playlist ranking.
    Technologies: Java, Perl, Python, JavaScript
  • Lead Software Development Engineer

    2006 - 2009
    • Managed the Search team, which generated about 40% of company revenue.
    • Led architecture as well as managed and mentored engineers.
    • Worked with the Live Search and MSR teams to incorporate Tellme technology into Microsoft Search and voice products.
    • Worked with speech and linguistics teams to dramatically increase automation of voice calls and relevance of text search queries.
    • Led design, development, and refinement of business category search.
    • Awarded a patent for a new spelling correction algorithm.
    • Developed algorithmic performance improvements that quadrupled back-end capacity while keeping relevance constant.
    Technologies: Perl, C, C++
  • Technical Lead

    2005 - 2006
    • Worked on contextual advertising, applying computational linguistics and machine learning techniques to the problem of finding relevant text ads to display on a webpage.
    • Developed an innovative classification algorithm to detect pages on which it would be inappropriate to display ads.
    Technologies: Perl, Python, C, C++


  • Nanomid (Development)

    Simple text-based music representation language with tools to convert to MIDI format.

    Suite of tools to manipulate MIDI files, such as automatically generating piano reductions of full scores, changing instruments, cropping, and more.


  • Languages

    SQL, Perl, Java, Python, JavaScript, C++, C
  • Frameworks

  • Paradigms

    Data Science
  • Other

    Software Developer, A/B Testing, Machine Learning, Analytics, Data Processing, Statistics, Mathematics, Data Analysis, Data Engineering, Data Modeling, Data Analyst, NLP, Data Engineer, Data Modeling Expert, Natural Language Processing (NLP)
  • Platforms

    Unix, Linux
  • Storage

    PostgreSQL, MySQL, Vertica


  • Master of Science degree in Computer Science
    2002 - 2004
    University of California - Santa Cruz
  • Master of Science degree in Mathematics
    2000 - 2002
    University of Illinois - Urbana-Champaign

To view more profiles

Join Toptal
Share it with others