Experience

SentiLink

• United States
• IT Services and IT Consulting
• 1 - 100 Employee

• Data Scientist

  • Sep 2022 - Present

• Data Scientist, Sales Support

  • Feb 2022 - Sep 2022



Warner Music Group

• United States
• Musicians
• 700 & Above Employee

• Data Science Intern

  • May 2021 - Feb 2022

  • Researched 1,000+ songs’ progression pattern across 12 territories and 500 playlists in Python, SQL, and Snowflake • Collaborated with the business department on quantifying marketing effect on songs online streaming and increasing songs streaming by an average of 77% on accurate marketing with R and time series modeling • Introduced multiple statistical models including Markov Chain to rank the importance of territory markets and increase the progression effects on songs… Show more • Researched 1,000+ songs’ progression pattern across 12 territories and 500 playlists in Python, SQL, and Snowflake • Collaborated with the business department on quantifying marketing effect on songs online streaming and increasing songs streaming by an average of 77% on accurate marketing with R and time series modeling • Introduced multiple statistical models including Markov Chain to rank the importance of territory markets and increase the progression effects on songs streaming by 10% • Developed a visualization tool to present progression behavior in Python with an interactive map in HTML Show less • Researched 1,000+ songs’ progression pattern across 12 territories and 500 playlists in Python, SQL, and Snowflake • Collaborated with the business department on quantifying marketing effect on songs online streaming and increasing songs streaming by an average of 77% on accurate marketing with R and time series modeling • Introduced multiple statistical models including Markov Chain to rank the importance of territory markets and increase the progression effects on songs… Show more • Researched 1,000+ songs’ progression pattern across 12 territories and 500 playlists in Python, SQL, and Snowflake • Collaborated with the business department on quantifying marketing effect on songs online streaming and increasing songs streaming by an average of 77% on accurate marketing with R and time series modeling • Introduced multiple statistical models including Markov Chain to rank the importance of territory markets and increase the progression effects on songs streaming by 10% • Developed a visualization tool to present progression behavior in Python with an interactive map in HTML Show less



Columbia University Irving Medical Center

• United States
• Hospitals and Health Care
• 700 & Above Employee

• Research Assistant

  • Oct 2020 - Feb 2022

  • Utilize NLP models including LSTM and BERT to classify and conduct sentiment analysis on medical students’ EPA (Entrustable Professional Activities) performance to support medical institutions’ evaluation. Achieved an average precision of 87% • Collaborated with Columbia Medical School to collect and process raw data, train on over 50,000+ records, incorporate deidentification and preprocess modules for hospitals and medical schools using Python • Deployed a web app based on… Show more • Utilize NLP models including LSTM and BERT to classify and conduct sentiment analysis on medical students’ EPA (Entrustable Professional Activities) performance to support medical institutions’ evaluation. Achieved an average precision of 87% • Collaborated with Columbia Medical School to collect and process raw data, train on over 50,000+ records, incorporate deidentification and preprocess modules for hospitals and medical schools using Python • Deployed a web app based on Flask to allow medical students to upload records and download analysis results. Classified 5000 records under 10 seconds compared to weeks of manual classification. Show less • Utilize NLP models including LSTM and BERT to classify and conduct sentiment analysis on medical students’ EPA (Entrustable Professional Activities) performance to support medical institutions’ evaluation. Achieved an average precision of 87% • Collaborated with Columbia Medical School to collect and process raw data, train on over 50,000+ records, incorporate deidentification and preprocess modules for hospitals and medical schools using Python • Deployed a web app based on… Show more • Utilize NLP models including LSTM and BERT to classify and conduct sentiment analysis on medical students’ EPA (Entrustable Professional Activities) performance to support medical institutions’ evaluation. Achieved an average precision of 87% • Collaborated with Columbia Medical School to collect and process raw data, train on over 50,000+ records, incorporate deidentification and preprocess modules for hospitals and medical schools using Python • Deployed a web app based on Flask to allow medical students to upload records and download analysis results. Classified 5000 records under 10 seconds compared to weeks of manual classification. Show less



Columbia University

• United States
• Higher Education
• 700 & Above Employee

• Teaching Assistant

  • Jan 2021 - Sep 2021

  Graded assignments, exams for graduate-level Algorithm (W4231: Analysis of Algorithm) for 130+ students. Topics covered: sorts, graph search, dynamic programming, greedy, NP-completeness. Hold weekly office hours.

• Research Assistant

  • Jan 2021 - Jun 2021

  • Conduced research under Project COSMOS, a cloud smart-city project funded by NSF, under the supervision of Prof. Zoran Kostic • Built a pipeline of neural networks in TensorFlow and Darknet for face/license plate detection and mosaic processing • Introduced and set up video annotation framework for research group and lead a team of 12 people in the task of generating a fineannotated dataset with 6,000+ images with CVAT and OpenCV • Improved the detection accuracy of the… Show more • Conduced research under Project COSMOS, a cloud smart-city project funded by NSF, under the supervision of Prof. Zoran Kostic • Built a pipeline of neural networks in TensorFlow and Darknet for face/license plate detection and mosaic processing • Introduced and set up video annotation framework for research group and lead a team of 12 people in the task of generating a fineannotated dataset with 6,000+ images with CVAT and OpenCV • Improved the detection accuracy of the pipeline by 20% by fine-tuning the networks and optimizing edge detection features on GCP



XuetangX

• China
• E-Learning Providers
• 1 - 100 Employee

• Data Engineer

  • Jun 2018 - Sep 2018

  Designed an advanced monitoring system for data processing and storing procedures, and proposed a new visualization toolkit for the aforementioned systems, with Numpy, Pandas, Scikit-learn, and so forth. Visualized system data of the past 3 years integrated with a real-time summary based on HTML webpage, and delivered weekly summary presentations of visualization in team meetings, in matplotlib. Designed an advanced monitoring system for data processing and storing procedures, and proposed a new visualization toolkit for the aforementioned systems, with Numpy, Pandas, Scikit-learn, and so forth. Visualized system data of the past 3 years integrated with a real-time summary based on HTML webpage, and delivered weekly summary presentations of visualization in team meetings, in matplotlib.