At Rackspace Technology (2022-current)

  • Developed Business Solution for Clients. Leveraged Public Cloud Platform (Azure, Databricks, AWS) to build Data Pipelines to ingest, transform, and consume data with Power BI reports. Designed and Implemented forecasting solution (Databricks, Prophet) for client’s operational costs.
  • Optimized the two most complex client’s data pipelines (Azure+Databricks) and reduce their total runtimes of them by 37.5% and 40%, respectively.
  • Built a production-grade chatbot application using Azure OpenAI that can enhance search capability; hence improved users’ efficiency. This leveraged a large amount of historical data.

At JD.com Silicon Valley Research Center (2017-2022)

  • Designed and implemented algorithms for advertising and marketing technology on JD platforms. Designed and analyzed experimental/observational data on user behaviors and for different ad products. Techniques: A/B testing, Experimental Design, Causal Inference.
  • Developed business growth strategies from data insights. Improved ~200M users and advertisers/brands experience, and contribute to increase in ads revenue.
  • Led personalized price prediction with advanced machine learning algorithms. Improved baseline accuracy by 20%. Designed production pipelines to push the trained models online and monitor model quality. Tools: JD’s proprietary task automation, mlflow.
  • Other Projects: Heterogeneous Treatment Effects Measurement, Experiment-Based User Recommendation, Omni-Channel Marketing Strategy Based on Offline Consumer Data and Location-Based Service.
  • Tools & Skills: PySpark, MLLib, Python, Scikit-learn, Keras/TensorFlow, R, SQL, Scala, Linux, HDFS.

At Northwestern University (2012-2017)

I am a PhD candidate in the Electrical Engineering and Computer Science (EECS) department at Northwestern University. I am a member of the Communications and Networking laboratory, working under the supervision of Professor Randall Berry (Northwestern University) and Professor Vijay Subramanian (now at University of Michigan).

I joined Northwestern University in September 2012 as a doctoral student under the Vietnam Education Foundation Fellowship. On the way, I earned my Master Degree here in June 2014 and continued through my PhD program.

My work at Northwestern includes studying bandwagon-type effects for models of Bayesian observational learning in social networks, which resembled many online recommendation systems (Amazon, Yelps, etc.). We studied models under different information structures and discovered many interesting, non-intuitive results: while noisy observations could help customers in their learning processes, additional information such as customers’ reviews/feedback could worsen their purchasing decisions. The main techniques used were hypothesis testing, Markov analysis, random walks, martingales processes and Monte-Carlo simulation. In addition, my research interest includes data science, machine learning, probabilistic models, game theory, quantitative finances, and optimization and resources allocation in communications networks.

Contact me at:

emails: thole2012[at]u.northwestern.edu, thonle2012@gmail.com