Kristi Kwon Email and Phone Number

• Designed, documented, and tested data logging for new API calls that could accomodate future product • Collaborated in cross-functional teams to define design requirements, prioritize resources, and write production codes in Python that calculates child performance in different areas of skills across 25 different games.• Applied feature engineering that increased Silhouette Coefficient by 14.2% in KMeans clustering of children in different stages of spelling.• Researched 4 different methods of modeling adaptive learning engine and created artificial activity data to simulate a learning path using RNN in Torch• Performed multiple weekly ad-hoc analyses in SQL for different stakeholders within the company, and presented insights on Tableau to VPs.• Created procedure that combined 11 data tables to increase readability and accesibility for other data analysts.

Metis is an immersive 12 week data science bootcamp in San Francisco with a curriculum designed by Datascope Analytics. Through five data projects, Metis teaches and reinforces concepts such as statistical inference, data acquisition, supervised and unsupervised learning, D3.js, SQL and NoSQL databases, and Hadoop MapReduce.Wine Uncorked: Predicting Price of Wine Jan 2017• Web scraped thousands of wines, weather, and annual US GDP data using BeautifulSoup and Selenium• Fine tuned parameters of gradient boost model using RandomizedSearchCV with a prediction error factor of 1.91• Analyzed feature importance through testing different values and visualizing its change in behaviorClassi-fly: Predicting Flight Status (Classification) Feb 2017• Trained one million domestic flight data on XG Boost Classifier model to predict the probabilities of departure and arrival flight status• Tackled imbalanced data through undersampling, feature engineering, and adjusting decision thresholds on performance metricsRecipe Recommender (Natural Language Processing)• Built recommender that interprets user input on any dish to recommend healthier alternative recipes from yummly• Utilized unsupervised learning techniques such as Latent Semantic Analysis (LSA), Principal Component Analysis (PCA), KMeans Clustering, and cosine similarities to identify similar yet healthy dishesPredicting Restaurant Closures (Classification, Natural Language Processing)• Identified time-variant features on Yelp reviews to differentiate open and closedrestaurants in San Francisco using sentiment analysis, Latent Semantic Analysis (LSA),and TF-IDF• Ensembled multiple classification tree models to best predict closures 3 months inadvance

• Manipulated and analyzed billion-lined datasets using SAS to produce workbooks and tools to monitor and evaluate healthcare policies surrounding End-Stage Renal Disease patients • Formulated relevant research questions, ran regression models using Stata, and conducted various ad-hoc investigations to evaluate the effect of Medicare programs • Designed interactive dashboards that monitor and interpret performance on a new government measure for over 6,500 dialysis facilities across the nation• Compiled findings, reviewed literature, and collaborated with clinical experts to interpret results and provide recommendations • Managed internal code repository that is used to produce project deliverables• Interviewed and trained new hires for onboarding project specific goals

• Imported and merged collected data into organized dataset to run regression models using Stata• Established a detailed documentation of variables and a record keeping system for data processing• Identified and resolved potential limitations through alternative statistical methodology• Analyzed and formed key insights from results to assist the completion of three research paper• Collected over 22.5 hours of data on the use of curb parking to evaluate influence of SFPark program

• Made judgment on appropriate location of the tanks for effective use at the Quivillaca school in Huatta, Peru based on research findings such as Unicef: WASH and the Sphere Project. • Calculated the volume of the first flush system to prevent contaminants built up on the roof, pipe sizing, drainage time, and number of pipes required for each building on which the system will be implemented.• Assisted in completion of writing the 524 Preliminary Design Report presented to EWB-USA, including CAD drawings of site plans and potential construction details

• Coordinated tabling by assigning responsibilities to other interns and preparing necessary materials to recruit and/or get petitions from students on campus• Convinced average of 18 students per hour to sign a petition to ban plastic bags statewide• Collected over 1500 petitions in a single semester and organized interest cards into the database for efficient communication with the public

• Researched and analyzed performance management and evaluation of information systems in various organizations• Drew implications from each case to provide a possible solution or new outline of IT service evaluation for National Information Society Agency (NIA) in Korea

• Discovered and analyzed EPC (Engineering, Procurement, and Construction) companies of their ownership structures and financials to evaluate the suitability of M&A for the client’s company• Forecasted financial outlook of the company by making inferences from data