Nov 2025 – Feb 2026 · 3 mos

Jul 2019 – Nov 2025 · 6 yrs 4 mos · San Francisco, California

Apr 2018 – Jun 2019 · 1 yr 2 mos · San Francisco, California

• · Built two models that predicts conversion and retention probabilities and used them to sort the feeding for different users and different platforms, so that the user conversion and retention rate is improved.
• · Built a siamese network to predict duplicates of newly uploaded products
• · Built a regression model to predict the hammer price for the auctioned products
• · Built recommender system for user personalization to increase sales
• · Built data pipelines with AWS and Spark to support the feature extraction, model applications

Jan 2016 – Apr 2018 · 2 yrs 3 mos · Washington D.C. Metro Area

• Major Projects:
• Built the win rate model with deep learning.
• Built the customer side market based pricing models.
• Individually Built machine learning models on customer and quotes/orders data to predict the win rate per customer per part quote.
• Corporate with Senior Data Scientist on merging two models (customer and vendor side) together to optimize the customer win rates and vendor taking rates.
• Did data analysis/insights to help management team to do better business decision, explained T-test, p-value, R squared and other statistical terminology well to non-professional people.
• Built the job matching/recommendation engine
• Corporated with CTO on building the recommendation engine
• Experimented predicting market based prices.
• Created vendor quality score function and blended the quality score in the job matching algorithm.
• One of the inventors of the engine team who built the auto quoting engine and won the patent for Xometry
• Built models to predict the machine set-up, cycle time and cnc_process classifier, and kept updating/retraining the models.
• Built several regression and classification models, experienced in data cleaning, preprocessing, feature extraction, model selection and tuning and model evaluation.
• Experienced in feature engineering and improved the models significantly
• Experimented different approach and used unvalidated data to achieve better models.
• Experienced in various python machine learning and scientific computing packages.
• Built the python/cython binding for HOOPS Exchange
• Managed to extracted valuable information from CAD files for better price prediction
• Took charge of Hoops maintenance, provided work-around/bug hotfix when Hoops has bugs.
• Worked with Hoops people in France for trouble shooting and debugging
• Geometry Analysis/ Feature Detection
• Created the tool direction algorithm
• Built speed and memory profiling tools for internal use

Sep 2015 – Dec 2015 · 3 mos · Baltimore, Maryland

• Vacant Housing Dynamics in Baltimore City Project Research Assistant
• Working with City officials, our goal is to better understand the dynamics of vacant housing in Baltimore City, measure the impact of current interventions, and help to hone decision and policy making using statistical analyses of available data.

Jul 2015 – Aug 2015 · 1 mo · Shanghai City, China

• Chinese Name Speech Recognition Engine
• Built the Chinese Name Recognition Engine from scratch individually using algorithms like max entropy and hiddem markov processes.
• Created the technical paper and created the probability formula based on Bayesian models.
• Calculated prior, posterior probabilities, tuned the confusion matrix
• Achieved better predictions than human recognition
• Data Processing
• Wrote python codes to help supervisors with data processing

Jun 2015 – Jul 2015 · 1 mo · Shanghai City, China

• focused on supporting market research analysis for various premium, luxury brand clients worldwide.
• Conducted customer clustering to achieve better marketing insights;
• Built regression models to see what effects the customer satisfaction most;
• Prepared qualitative and quantitative analysis from worldwide surveys of missions assigned;
• Extracted data and key findings from the database of related market surveys;
• help implement new templates for reports and questionnaires based on client’ s requirements;

Jan 2015 – May 2015 · 4 mos · Baltimore, Maryland

• Yuan is a teaching assistant of the 550.310 Probability & Statistics for the Physical &Information Sciences & Engineering