Help & Supportcareers@gravityer.com
AI is analyzing your overall score…
Identifying your key strengths…
Evaluating your skill match against the job requirements…
Assessing your cultural and operational fit
Engineering Lead at Instagram | TikTok | ex-Facebook
I am an engineering tech lead at Instagram Threads team leading new/low active user recommendation team and LLM for content quality team. Previously I was TL in TikTok feed recommendation team with a focus on ML projects of new user growth. Before 2022.6, I was engineering tech lead (15-20 people) at TikTok ads targeting/signal/measurement team with a focus on ML projects. Before joining TikTok, I was a research scientist/tech lead in app ads ranking (with xxB annual revenue) at Facebook, with rich experience building and improving e2e ads machine learning products and systems. I Previously worked as machine learning researcher at BMW Technology Corporation with 6 patents. I got my Ph.D. in Intelligent Transportation, and M.S. in Computer Science from UW-Madison with strong programming background and machine learning/data mining skills to draw insights from data.
University of Wisconsin-Madison
Master's Degree, Computer Science
January 1, 2014 – January 1, 2016
University of Wisconsin-Madison
Doctor of Philosophy (Ph.D.), Intelligent Transportation Systems, Civil Engineering
January 1, 2013 – May 1, 2017
Xi'an Jiaotong University
Bachelor's degree, Electrical, Electronics and Communications Engineering (with highest honor)
January 1, 2009 – January 1, 2013
Research Scientist, Tech Lead in Threads
September 1, 2024 – Present
Seattle, Washington, United States · Remote
TikTok
Engineering Lead - New User Growth/Quality Staytime/Feed Quality
June 1, 2022 – August 1, 2024
Seattle, Washington, United States · Hybrid
ByteDance
Engineering Lead - Ads Targeting/Identity/Signal Modeling
June 1, 2020 – June 1, 2022
Seattle, Washington, United States
Research Scientist, Tech Lead in App Ads Ranking
August 1, 2018 – June 1, 2020
Greater Seattle Area
BMW of North America, LLC
Machine Learning Researcher
January 1, 2017 – July 1, 2018
Chicago
Traffic Operations and Safety Lab, University of Wisconsin
Project Assistant (Web Developer)
September 1, 2013 – January 1, 2017
Traffic Operations and Safety Lab, University of Wisconsin
Graduate Research Assistant
September 1, 2013 – January 1, 2017
Systems Engineering Institute, Xi'an Jiaotong University
Research Assistant
March 1, 2013 – June 1, 2013
Xi‘an, Shaanxi, China
Smiles (Social Mobile Multimedia Mining Learning & Search Group) Lab, Xi'an Jiaotong University
Research Assistant
September 1, 2011 – July 1, 2013
Xi'an, Shaanxi, China
Systems Engineering Institute, Xi'an Jiaotong University
Research Assistant
May 1, 2011 – August 1, 2011
Xi‘an, Shaanxi, China
Mathematics and Statistics, Xi'an Jiaotong University
Visiting Student
April 1, 2010 – February 1, 2012
Xi'an
Entity Matching for Electronics Products from Amazon and BestBuy
December 1, 2015 – Present
In this project, electronic products is chosen as our study domain. We select two Web sources (one from Amazon, the other from BestBuy), crawl to retrieve HTML data, perform information extraction to convert the HTML data into two relational tables. Next, we use Magellan, a data matching system develped at Wisconsin, to do the blocking and matching for the two tables.
Traffic Congestion Classification Using Cellular Phone Activity Data
December 1, 2015 – Present
Firstly, precisions using support vector machine with different kernel functions and neural network with different hidden units and output functions are compared by 3-fold cross validation. Next, bagging (Bootstrap Aggregation) algorithm is introduced with the SVM and Neural Network chosen by first step, and confusion matrix among three different congestion levels is used to evaluate algorithms. In the last step, we also investigate how predictive accuracy varies as a function of training-set size. Experiments show that ensemble methods outperform single classifiers, and Bagging-SVM with RBF kernel outperforms all other classifiers for this classification task
Small Version of Yelp
May 1, 2015 – Present
We used Oracle 11g as our relational database. Struts including jsp, java, xml is used as our web application MVC. We build an small version of Yelp which enables users to search nearby restaurants, events, add/delete friends, add/change comments and ratings,
Movie Recommendation System for Group Audience
December 1, 2014 – Present
Movie recommendation systems provide movie recommendations for the audiences according to their rating on several movies they have watched. These systems match the audience’s preference with movies in a large pool and pick movies closer to the audience's taste. There is a set of movie recommendation systems proposed in the market. However, most of them consider only the taste of individual movie watcher. If the audiences intend to watch movies in a group, these systems could not provide useful recommendations. In this study, we design a group recommendation system. An interview test is conducted to collect the movie preference inputs from different gender combinations of group audiences (male-male, male-female and female-female). By considering preferences of all the audiences in the group, this system applies the item-based collaborative filtering algorithm and gives movie recommendations for the group movie watchers. The group audiences score the recommendations from the proposed system as well as the ones produced by the traditional individual recommendation system. The scores of the two systems are compared to measure users’ preference. The results show that the group recommendation system gets a higher score from the group audience than the individual recommendation system. The proposed system provides a better recommendation for group movie audiences.
No More Wait for River Trip
February 1, 2012 – Present
To allow more people to enjoy the exciting water rafting in the Big Long River, we develop detailed schedules for different types of trips, deciding launch time and travel agenda based on multi-objective optimization. This work got Meritorious Winner of Mathematical Contest In Modeling.
Modeling Electricity Pricing through Multi-objective Programming
January 1, 2012 – Present
Considering the effects of energy efficiency and energy consumption on electricity pricing, a multi-objective programming model is constructed and the maximal economic benefit and minimal environmental pollution are taken as the objective functions. Co-integration analysis is performed for modelingthe consumer behavior, and cost-profit model is chosen to measure the producer behavior, which serve as the constraint conditions of multi-objective programming model.
Electric Vehicles, Just Sounds Environmental and Economical?
February 1, 2011 – Present
After mathematically analyzing and modeling EV problem, we draw a conclusion that not only the EVs are environmental friendly and economical, but also the widespread use of them is feasible and practical. This work got Meritorious Winner of Interdisciplinary Contest In Modeling.
Cultural Fit Analysis
The candidate's career trajectory shows a strong drive for impact and innovation, consistently taking on leadership roles in high-growth, data-intensive environments. The diversity of projects, from academic modeling contests to leading AI initiatives at Instagram and TikTok, demonstrates adaptability and a broad interest in applying advanced analytical techniques to real-world problems. This aligns well with a culture that values continuous learning, problem-solving, and leadership in data-driven domains. The target role of 'Data Analyst' might be a slight mismatch given the candidate's extensive experience as an Engineering Lead/Research Scientist in Machine Learning/AI; they appear overqualified for a typical Data Analyst role and would likely seek more senior, strategic, or ML-focused positions.
Soft Skills & Operational Fit
The candidate's experience as a Tech Lead at multiple major tech companies indicates strong leadership, team-building, and cross-functional collaboration skills. The descriptions of leading teams from '0 -> 5-10' and driving 'over 10 cross org XFN teams' suggest excellent operational fit for roles requiring significant influence and project management. The academic background and project work also highlight strong problem-solving and analytical capabilities.