Adrian Cheung Email & Phone Number

South San Francisco, Ca, Us

Ann Arbor, Michigan, Us

- Designed a Classifier Evaluation Pipeline to streamline ML experiments and result analyses- Led the effort to implement a multi-intent feature in the flagship custom AI assistant platform- Improved the Slot Mapping testing infrastructure runtime from an order of hours to minutes- Investigated/implemented various semi-automated clustering approaches to reduce the manual effort for sample utterance analysis by 95%- Mentored team, interns, and junior engineers on effective testing, productivity, feedback, etc.- Contributed to the publication of two papers on data curation by conducting literature searches, designing/running experiments, data analyses, paper writing

Knowledge Graph Team- Developed a system for hand tuned decisions trees for entity deduplication across multiple providers- Experimented with Machine Reading Comprehension as a means of attribute expansion- Worked on data ingestion pipelines ranging from JSON, TSV, XML, and other custom markups- Managed an intern; designed their project, mentored them to a successful internship with a full time offer Objection Content Classification Team- Developed features to train a machine learned video classifier for nudity and pornography- Assisted with integration of a video segment labeler tool used for collecting training data- Implemented Facebook’s first live video classifier for violence which had a higher action rate and fewer false positives than human reports

- Implemented a semi-exhaustive search for improved ad placement in Facebook’s Newsfeed- Performed speed and improvement analysis on newly added auction algorithm- Summarized all work and analysis into a final report to aid future company decisions

San Francisco, Ca, Us

- Categorized functions of mechanical components using Word2Vec and pre-tokenized text- Applied deep neural networks using Caffe to classify fuel consumptions of cars based on side profile images- Explored multi-labeling techniques with machine learning for feature identification of furniture products- Data extraction from multiple sources including JSON, CSV, and SQL databases

- Implemented a Part Of Speech Tagger in C++ using Brill Tagging rules- Performed speed and accuracy tests for the part of speech tagger- Added new features to improve content ranking for searching of posts- Ran experiments on newly added features and verified results showed statistical improvements

Chicago, Il, Us

- Implemented front end features for Richtree’s online ordering using Javascript and AngularJS- Introduced new logic behind an online registration form for a Rails based web application- Implemented a web data scraper and corresponding API using Rails and Capybara- Introduced a feature to identify potential duplicate entries in a mySQL database based on similarity over multiple properties- Actively used Test Driven Development for all feature work using Rspec, Jasmine.js, and Cucumber/Capybara

Menlo Park, Ca, Us

- Worked under an Agile team environment with daily stand-up meetings- Developed front end features for various call-center applications- Actively used Test Driven Development for all features- Configured automated testing and deployment for a continuous integration environment- Technologies used: Javascript, HTML5, CSS, Backbone.js, Bootstrap, Node

Toronto, Ontario, Ca

- Worked under an Agile team environment with daily stand-up meetings- Updated content on Rogers.com with HTML and direct image uploads- Ensured changes to the website were accurate and user friendly- Provided insight on overall user experience and recommended enhancements- Wrote a guide outlining the steps for a newly integrated content management system

Bachelor Of Software Engineering With Psychology Minor

International Baccalaureate Certificate For Mathematics And French