My name is Binbin Li. I am currently a Data Science Manager at Facebook, also known as Meta. I am with the Facebook Ads Backend & Systems Analytics team in the Ads & Business Products org. I lead / have led mutliple data science teams, including

  • Ads Personalization DS team in Ads Core ML: feature & data for ~300 ads ranking models across Meta ecosystem ads surfaces (Facebook blue app, news feed, Instagram etc.)
  • Ads ML Infrafoudation DS team: feature (GSF/GFF/F3) and model traing platforms (DPer3/Pytorch)
  • Ads ML Automation DS team: MasterCook/LineCook model development processess and AutoML/Blueprint platforms
  • Ads ML Privacy Experimentaton: A/B test also internally known as QRT under multiple signal loss scenarios such as App Tracking Transparency (ATT), Online Behavioral Advertising (OBA), ePrivacy Directive (ePD) etc.

Previous Experience

Before Facebook, I was Head of Data Science at true[X], a technology company building the next generation of advertising products and experiences for premium video. Ranked as one of the world’s most innovative companies, true[X] was a subsidiary of the formerly 21st Century Fox, and later part of The Walt Disney Company. During my time with true[X] / Disney, I led a team of very talented data scientists, machine learning engineers and software engineers to build:

  • UP//LIFT Optimize, an intelligent ad decisioning engine powered by machine learning models that assess how consumers are reacting to a brand in real time, match the right consumers to the right ads, and optimize performance of ad campaigns running on true[X]’s ad platform. Independent study demonstrates that our machine learning models “dramatically increase brand familiarity, brand interest, and purchase consideration”.

  • UP//LIFT Monitor, an industry-leading online survey sampling and collection system that correctly and precisely measures performance of ad campaigns on and off true[X]’s ad networks.

Before true[X], I worked for KPMG US as a Data Scientist Manager at KPMG Lighthouse - Center of Excellency in Data & Analytics, where I delivered novel machine learning and artificial intelligence models to help clients turn data into value. Clients included Freddie Mac and Johnson & Johnson.

Before KPMG, I worked at SAS Instituite and led the core data science team of SAS Fraud Management Solution, the industry leading real-time fraud management system to combat credit/debit card and online banking fraud. We had built real time fraud detection models for major top-tier banks all over the world, and helped clients prevent fraud loss of hundreds of millions dollars. If you live in the US and have a credit/debit card, chances are the neural networks models we build are/were pretecting you and your wealth from fraud.

Before SAS, I interned at the Market Risk Modeling Group of Goldman Sachs, where I developed a new algorithm up and running in prouduction to generate large dimensional, highly correlated time series for missing market data and to recover real data correlations.

Education

I obtained my Ph.D. in Systems Engineering (a multidisciplinary research field particularly on Optimization, Control, Operations Research, and Machine Learning), and M.Sc. in Electrical Engineering at Boston University, advised by Prof. Yannis Paschalidis. I also hold a M.Sc. degree in Control Science and Engineering, and a B.Sc. degree in Automation from Tsinghua University, Beijing, China, working with Prof. Ling Wang. I have broad research interests in the areas of Optimization, Control, Operations Research, Evolutionary Computation, Machine Learning (including Deep Learning, Reinforcement Learning / Neuro-Dynamic Programming), and Artificial Intelligence. I have published four IEEE Transactions journal papers, four academic society flagship conference papers, one book chapter, and filed two patents, all in the aforementioned research areas.

Technical Competencies

  • Programming Languages: Python, Scala, Perl, C, C++, Shell Script, SQL
  • Statistical & Optimization Languages: SAS, R, ILOG CPLEX, MATLAB, ATLAS/LINPACK
  • Cloud Platform & Big Data: AWS, GCP, Databricks, Apache Open Source Software (Spark, Hadoop, Hive, Flink etc.)
  • Machine Learning: Statistical Learning, Deep Learning, Reinforcement Learning / Neuro-Dynamic Programming, PyTorch, TensorFlow, XGBoost, LightGBM, scikit-learn, Spark MLlib etc.