Experience

Kinship AI

• Consumer Services
• 1 - 100 Employee

• Co-Founder

  • Oct 2022 - 1 year 3 months

  San Francisco, California, United States We design digital siblings to be there when you need and thinking of you when you don't. Beating with the heart of their builders, they don't care who or what they should be, only what you need them to be.



Neeva

• United States
• Technology, Information and Internet
• 1 - 100 Employee

• Software Engineer - Machine Learning

  • Nov 2020 - Oct 2022

  San Francisco Bay Area * trained models for UGC (user generated content) ranking * identified when to trigger UGC content * Built the eval and metrics pipeline used by the Web Search team * trained and iterated on the search ranking model * introduced click data as a source of training & eval data * worked with external vendors to generate ground truth data



Etsy

• United States
• Software Development
• 700 & Above Employee

• Staff Machine Learning Engineer

  • Oct 2020 - Nov 2020

  San Francisco, California, United States * Lead a team of 5 MLEs focused on Machine Learning Systems * Built a feature store to support multiple ML teams * Added vector based features to the search ranking models

• Senior Machine Learning Engineer

  • Oct 2018 - Oct 2020

  San Francisco, California * Tech lead on Machine Learning Systems (we integrated an ANN into the production stack) * Worked on an evolutionary strategy framework to balance the tradeoffs between multiple optimization goals: https://github.com/etsy/Evokit



Google

• United States
• Software Development
• 700 & Above Employee

• Software Engineer

  • Aug 2016 - Aug 2018

  Mountain View, California ● Validated a new proxy label for user activity by identifying and correcting bugs in the data collection pipeline. Built a dashboard to monitor its distribution at collection time to avoid training on corrupt data. ● Designed and implemented many new features for measuring users’ ad activity on Youtube that were validated and deployed across multiple models. Many of these features were generalizable to other teams. ● Worked with Data Scientists and Product Managers to frame ad… Show more ● Validated a new proxy label for user activity by identifying and correcting bugs in the data collection pipeline. Built a dashboard to monitor its distribution at collection time to avoid training on corrupt data. ● Designed and implemented many new features for measuring users’ ad activity on Youtube that were validated and deployed across multiple models. Many of these features were generalizable to other teams. ● Worked with Data Scientists and Product Managers to frame ad watch time within our business context and designed solutions to accurately model the relevant portions of its distribution. Implemented the backend changes to support composing different model predictions in our online systems. ● Performed feasibility analysis for new features & model architectures. Rapidly prototyped different designs on data sets ranging from 100M-10B+ examples. Tested and validated candidates in A/B experiments. Deployed models used in the Youtube Ads auction to predict watch time, ad abandons, and other user activity. Continuously worked to improve their respective R-squared and AUC. ● Developed and pitched an idea for collecting new user interaction signals that aims to improve user ad relevance. Worked with product, design, frontend and legal to synthesize clear requirements for their implementation. I am currently leading this cross-functional project. ● Member of on call rotation and responded to online outages. ● Mentored an intern and oversaw his work on new feature development and model re-architecture. Show less



Cornell University

• United States
• Higher Education
• 700 & Above Employee

• Machine Learning Teaching Assistant

  • Aug 2015 - Dec 2015

  Ithaca, New York Area Hold office hours Grade homeworks and exams Monitor Piazza and answer questions Scribe for lecture notes Update projects



Cornell Data Science

• United States
• Information Services
• 1 - 100 Employee

• Education Lead

  • Dec 2014 - Dec 2015

  Ithaca, New York Organize a 5 hour crash course for members on Data Science Hold weekly educational sessions Attend e-board meetings and assist in club development



Numenta

• United States
• Software Development
• 1 - 100 Employee

• Algorithms Intern

  • May 2015 - Aug 2015

  Redwood City, CA Allowed users to create custom C++ and Python regions with nupic Made a region for pooling temporal data called a Union Pooler Worked on the NLP framework and created multiple models to test our performance Curated text data and designed a hierarchy Updated the Python build and pipeline for nupic so that we could eventually support other languages and Windows Analyzed seismic data for early earthquake prediction Created a phonology based encoder for letters that could later be… Show more Allowed users to create custom C++ and Python regions with nupic Made a region for pooling temporal data called a Union Pooler Worked on the NLP framework and created multiple models to test our performance Curated text data and designed a hierarchy Updated the Python build and pipeline for nupic so that we could eventually support other languages and Windows Analyzed seismic data for early earthquake prediction Created a phonology based encoder for letters that could later be used for speech synthesis Show less



Cornell University

• United States
• Higher Education
• 700 & Above Employee

• Computer Vision Teaching Assistant

  • Jan 2015 - May 2015

  Ithaca, New York Hold office hours Grade homeworks and exams Monitor Piazza and answer questions Ported the projects from C++ to Python



Blackbird Technologies, Inc

• United States
• Technology, Information and Internet

• Artificial Intelligence Intern

  • May 2014 - Aug 2014

  Menlo Park, CA Worked on detecting the dominant color in an image using a decision tree and an SVM Implemented a spatial pyramid to improve image classification Implemented soft k-means to handle ambiguity in image classes Created an auto-cropper to crop rotated images Worked on polka dot and pattern detection in images Incorporated PCA into the system Worked with NLTK to make parts of speech tagger Modified the hashing to account for the bias caused by hashing collisions Read many… Show more Worked on detecting the dominant color in an image using a decision tree and an SVM Implemented a spatial pyramid to improve image classification Implemented soft k-means to handle ambiguity in image classes Created an auto-cropper to crop rotated images Worked on polka dot and pattern detection in images Incorporated PCA into the system Worked with NLTK to make parts of speech tagger Modified the hashing to account for the bias caused by hashing collisions Read many research papers on computer vision Classified images and query logs Tested the classifiers Worked with SVMs Used bash scripts and Python with OpenCV Show less



Cornell University

• United States
• Higher Education
• 700 & Above Employee

• Calculus Course Assistant

  • Aug 2013 - Dec 2013

  Ithaca, New York Graded homework Held study sessions and helped students in the class with their work



AWP (Ancient Wisdom Productions Inc.)

• United States
• Design Services
• 1 - 100 Employee

• iOS Programming Intern

  • Apr 2013 - Dec 2013

  Ithaca, New York Added new features to a pre-existing application Piction Created an iOS application from scratch



AnthroTronix

• Research
• 1 - 100 Employee

• Software Engineering Intern

  • May 2013 - Aug 2013

  Silver Spring, MD Learned the basics of Android app development and made a basic application Work with Matlab for data analysis using graphs and number manipulation Computer Vision using Python and OpenCV to detect objects in images (NDA) Made a game called Tap Tap Rehabilitation to use with a stroke glove in Java Used Processing to create a version of Tetris to use with a Makey Makey, a device that can make any conductive material into an input Wired an Arduino and programmed a GUI to change the… Show more Learned the basics of Android app development and made a basic application Work with Matlab for data analysis using graphs and number manipulation Computer Vision using Python and OpenCV to detect objects in images (NDA) Made a game called Tap Tap Rehabilitation to use with a stroke glove in Java Used Processing to create a version of Tetris to use with a Makey Makey, a device that can make any conductive material into an input Wired an Arduino and programmed a GUI to change the color of an LED board connected to the microcontroller in Java Software testing Show less