Jason Katz Email & Phone Number

New York, Ny, Us

Sunnyvale, Ca, Us

Part of the Marketplace AI team, working to improve job seeker experience through the improvement of Job Search and Job Recommendation (JYMBII) algorithms• Reduced runtime of data cleanup flow by 95% through parallelizing Hadoop commands to run in batch, preventing silent failures due to timeouts• Added job title expansion using economic graph-based features to the JYMBII query model. The features are particularly noteworthy, as the economic graph is one of LinkedIn’s main competitive advantages over other job sites such as Indeed and this was the first time anybody had leveraged them in these models.

Sunnyvale, Ca, Us

• Created metrics to approximate facepalm results using Spark Scala• Utilized UMP and SQL to generate metrics to evaluate guest job search results• Trained offline Tensorflow models and deployed online, conducting A/B Tests• Developed pipeline to generate Job Activity Features

San Francisco, California, Us

I work as a freelancer for hire for different data science and machine learning projects, here are a few projects I have performed for clients in the past:• Built models to predict stock prices to algorithmically trade futures• Built a bot to delete all prior posts from a company's page on LinkedIn• Used data from TV commercial airings to model web attribution• Scraped luxury watch websites to collect data for real-time wholesale prices• Built a classifier to identify fraudulent transactions• Developed a model to predict the chance for a couple to get divorced using census data• Utilized gradient boosting methods to predict the strength of a patent• Built a model to estimate the value of a home along with its comparables

Los Angeles, California, Us

• Implemented fuzzy matching algorithm to cluster similar real estate buyers into single entities• Created algorithm to rank the relevance of a buyer to a potential property• Hosted server to enable HTTPS requests for efficient website integration

Providence, Rhode Island, Us

• Built a Julia package to implement the CAOS algorithm, which classifies diseases/organisms through phylogenetic analysis of DNA sequences• Utilized ICU data to predict potential patient conditions based off of which medications they were taking• Utilized metaprogramming to launch hundreds of jobs in parallel on Brown's supercomputers, performing data transformations and classifications on approximately 100 GB of data as part of research into the accuracy of the CAOS algorithm• Built pipeline from PubMed to Julia using SQL to execute UMLS queries• Performed association rule learning to find correlation among akin diseases of Adult and Pediatric Asthma from medical terms in articles

• Led daily standup meetings for a team of overseas developers guiding them to create a system to collect and store user data from the Papr app• Utilized IBM Watson's Natural Language Understanding to extract topics from news article headlines and descriptions• Built a model to recommend articles to users based on their historical interests

• Used data from court cases that challenged patents to build a model to predict the strength of a patent• Created a script to extract data on millions of patents from patentsview.org• Used historical patent data to assign strength scores to over 6 million active patents to be used for company evaluations

Arlington, Va, Us

• Part of a team of Brown University Computer Scientists who received a $3.1 million grant from DARPA to build an interactive data exploration system that includes statistical safeguards against false discoveries. • Built an interactive tool to perform system diagnostics on the machine learning optimizer • Performed analysis of thousands of datasets and millions of runs extracted from OpenML to assess the accuracy of different sklearn and weka primitives and create dataset clusters • Automated the process of cleaning data and building machine learning models, using results from OpenML analysis to automatically classify a dataset into a pre-defined cluster, narrowing down which methods are tested when fitting a model on the data • Built various machine learning models using Scikit-Learn to classify a question/sentence pair as either an answer or not

Miami, Florida, Us

• Utilized Pandas to perform data analysis on millions of rows of call data• Created visualizations using Python to provide insights into advertisement response patterns• Built a web attribution model to predict web response to advertisements• Used sklearn to build a gradient boosting model to predict what percentage of a clients' requested advertisement budget would clear

Sunnyvale, Ca, Us

• Performed feature engineering, extracting data through Pig scripts• Trained an offline model in Python to autocomplete job search queries • Used Java to implement the trained model for online deployment• Conducted A/B testing to validate model performance

Sunnyvale, Ca, Us

• Built an internal tool written in Java, to allow easy onboarding of Spark jobs for LinkedIn engineers• Automated the execution of ETL logic by extracting data from HDFS and connecting to Spark clusters• Translated scripts from Pig to Spark-SQL and Spark-Scala• Used Hive to provide validation of datasets by generating queries for output tables

Lincoln, Ne, Us

• Improved accuracy of multi-object tracking algorithm by optimizing parameters through grid and cube searches• Utilized AWS Batch to run large scale experiments via Boto3• Built a Bokeh App to simulate incoming workload for Hudl's team of sports analysts leading to changes in staffing• Used Python to build an expected goals model based off player tracking and event data from soccer matches• Performed sentiment analysis of support emails to gain insights into common customer frustrations

TurnUp is a mobile application that helps college students and young adults find popular events at local nightclubs and bars. I was part of the team that launched the app in March 2016 on the iOS App Store. • •Performed quality control to find and remove bugs from the app prior to the March 2016 launch• •Using a budget of $500, devised and implemented a marketing campaign; designed and led the team that distributed flyers, resulting in 2,500 downloads within 1 month of launch• •Coordinated with lawyers to draw up legal documents for the formation of the company•• Met and negotiated with investors to close a deal to raise $15,000 in capital for TurnUp in exchange for equity in the company• •Created the company budget and managed and documented revenues and expenses using Microsoft Office

Master'S Degree, Data Science

Bachelor Of Science (B.Sc.), Statistics

Data Science

Bachelor Of Arts (A.B.), Applied Mathematics