Machine Learning Scientist
Palo Alto, California
Staff Software Engineer @ Facebook Core Machine Learning, Ads Ranking.Senator, we run ads. From March 2018 to July 2019 (1 year 5 months) San Francisco Bay AreaSenior Machine Learning Scientist @ A9.com Our team owns Product Search ranking of Amazon.com sites worldwide. Based on my dual expertise in Engineering and Machine Learning, I have contributed to the...
Staff Software Engineer @ Facebook Core Machine Learning, Ads Ranking.Senator, we run ads. From March 2018 to July 2019 (1 year 5 months) San Francisco Bay AreaSenior Machine Learning Scientist @ A9.com Our team owns Product Search ranking of Amazon.com sites worldwide. Based on my dual expertise in Engineering and Machine Learning, I have contributed to the following areas:Machine Learning:- Researched, designed and implemented (C++) parallelization training of our ranking models. The new approach made models training >6x times faster and enabled the use of larger training sets. It was quickly adopted as the default of the Relevance team for modeling.- Designed the foundations of our "engagement-aware" ranking to model long-term customer satisfaction. I researched customer behavior post-search, implemented an attribution model for this feedback, created new objective functions and ranking features to model it, implemented and A/B tested the first few ranking models in several categories. The impact of launched changes is >$500M/yr. Now leading efforts to extent and iterate on this with the broader team.Online Serving:- Led the development on the Relevance side for a new deployment pipeline for ranking changes. I collaborated with top talent selected across the Search org, contributing to the design and its C++ code. The new deployment pipeline enabled considerable gains in productivity via simple, daily, and safe deployments, replacing a slow weekly release cycle. - Designed and implemented our text-pipeline segmentation algorithm for Chinese. This Machine Learning module is called for every query and document on Amazon.cn.Feature Engineering:- Designed and implemented the top ranking feature powering the "Sort by Featured" rank on Amazon websites worldwide. This ranking is the default for keyword-less searches, a sizable portion of the site's traffic.Offline Data Processing:- Mentored and worked with my intern to optimize a full join and its related logic on >100 terabytes data on Hadoop.- Recognized expert in debugging and fixing our sophisticated data pipeline on terabytes data, frequently solicited by management on high stakes deep dives. From January 2012 to March 2018 (6 years 3 months) San Francisco Bay AreaResearch Engineer @ Alibaba Group Worked at Alibaba Group R&D center to support its subsidiaries, including Taobao, Alibaba, Alimama and etc.- Tech lead of a 5 people team to develop a widely used text classification platform (in production for the Taobao contextual match advertisement system and the Alibaba's web search engine). Responsible for design, coding, modeling and project management. - High throughput, handle millions of requests per day with
Staff Software Engineer
March 2018 to July 2019
San Francisco Bay Area
A9.com
Senior Machine Learning Scientist
January 2012 to March 2018
San Francisco Bay Area
Alibaba Group
Research Engineer
June 2008 to February 2010
Yahoo!
Software Development Engineer II
March 2010 to January 2012
What company does Sheng Peng work for?
Sheng Peng works for Facebook
What is Sheng Peng's role at Facebook?
Sheng Peng is Staff Software Engineer
What industry does Sheng Peng work in?
Sheng Peng works in the Computer Software industry.
Who are Sheng Peng's colleagues?
Sheng Peng's colleagues are Yousuf Haque, Rafal Szymanski, Neha Sherpa, Alejandro Martinez, Yoojin Ivanovic, Jerry Su, Padigender Reddy, Austin Hulse, Jianyang Zhao, and André Silva
Enjoy unlimited access and discover candidates outside of LinkedIn
One billion email addresses and counting
Everything you need to engage with more prospects.
ContactOut is used by
76% of Fortune 500 companies
Sheng Peng's Social Media Links
/company/f... /school/ue...