Tianqi (Elena) Shen
email@example.com 213-***-**** Open to relocate
New Grad Software Engineer with a strong data background Education
University of Michigan Ann Arbor, MI Sep. 2018 - May 2020
• M.S. in Information, Data Science track (GPA: 3.98/4.0)
• Courses include:
- Data Science: Data Manipulation and Analysis, Information Visualization, Information Retrieval, Data Mining, Applied Machine Learning
- Web Full-stack: Database Application Design, Design of Complex Websites, Mobile Development Kalamazoo College Kalamazoo, MI Sep. 2014 - Jun. 2018
• B.A. in Economics & B.A. in Mathematics (GPA: 3.61/4.0), Honor: Cum Laude Skills
Frameworks: Django, ReactJS (React Native), Pandas, scikit-learn, matplotlib, Flask, d3.js Platforms & Tools: Git, Firebase, Jupyter, Tableau, Spark, AWS (EMR) Skill Sets: Full-stack Web Development, Data Mining, Machine Learning Work Experience
Software Engineer (position eliminated), ExtraHop Seattle, WA
• Due to COVID-19 position was eliminated.
Software Developer Intern, University of Michigan Ann Arbor, MI Jun. 2019 - Aug. 2019
• Launched Landing, Degree Detail and Partner pages on Michigan Online, a U-Michigan version of Coursera, which increased discoverability of upcoming online degrees and courses.
• Completed backend development of new features using Django and PostgreSQL, including email subscription service using Mailchimp APIs.
• Collaborated cross-functionally with UX engineers to drive frontend integrations, improve mock designs and provide readability consultation on Vue.js and CSS (LESS).
• Mastered sprint planning and tracking for two weeks, aligned tasks between engineers and UX and delegated actionable items to keep the team unblocked.
Instructional Aide, University of Michigan Ann Arbor, MI Jan. 2020 - Apr. 2020
• Held office hours of SI507 Intermediate Programming in Python, wrote automated tests to grade assignments and projects.
• Helped preparation of instructional course materials, including: object-oriented design, unit testing, API fetching, tiered caching, SQL, Flask, etc.
Prediction of Coronavirus Disease Dynamic Ann Arbor, MI Jan. 2020 - Apr. 2020
• Proved that timely and extensive detection would help to contain the spread of COVID-19; replayed the trend of confirmed cases in Wuhan and Hubei province to deduce a credible range of day zero of this pandemic.
• Tuned a LSTM model that can project month-long trends of confirmed cases with training data of merely the first week of outbreak, which foresees inflection points in all Chinese cities and countries such as Korea, reaching a mean nRMSE of 0.116 across 28 cities in cross-validation. Tone Classifier for Online News Ann Arbor, MI Oct. 2019 - Dec. 2019
• Built an interactive web console using Flask in Python and Bootstrap, which could be released as a public tool for distinguishing biased, objective or pseudoscience contents from daily news.
• Conducted text vectorization and semantic extraction in data-preprocessing, using Word2Vector and TF-IDF indexing utilities from NLP toolkits (NLTK and scikit-learn).
• Designed a multi-class news tone classifier with unbalanced data handling; compared outcomes among various models including SVM, LightGBM, CNN, and landed on an ensemble model producing an optimal F1-score of 0.82.
ACM CHI Conference 2019 - Shadoji Mobile App Development Ann Arbor, MI Oct. 2019 - Dec. 2019
• Partnered with UX designers on their ACM CHI 2019 Conference Design Competition finalist project, an app for those suffering from identical beauty standards, which aims to promote body shape diversity.
• Implemented the prototype of an Instagram-style mobile app using React Native and Firebase; users will be able to post, share their shadojis and comment on others’ shadojis. Personalized Yelp Recommender System Ann Arbor, MI Oct. 2019 - Dec. 2019
• Presented a cognitive filtering based Yelp recommendation system performing at a RMSE of 0.806 on the test dataset, which adopts six refined features that are reduced from 65 original dimensions by feature engineering.
• Exercised feature extraction and hyper-parameter tuning on the 9.8GB Yelp dataset, evaluated models such as Ridge, Polynomial Lasso and RNN, and chose a Random Forest regressor per optimal RMSE.