Alexander Pan Email & Phone Number

New York, Ny, Us

● Executed decisions related to data processes, management, visualizations, accuracy, automation, and more● Optimized BigQuery to ensure efficient SQL querying over large sets of complex data in combination with Looker● Created the Looker views and models necessary to run our BigQuery data infrastructure to work in tandem with theengineering team’s embedding process● Processed data into actionable insights for Network Next thru Looker data visualizations● Developed customer-facing dashboard visualizations to allow game studios to make informed business decisions● Trained, validated, and built machine learning models to provide insights, like feature importance and implicitfeedback, to investigate gaming issues for game studios, such as reported matches in relation to network performance● Determined network relay performance in order to find under-performing relays to shutdown and reduce costs● Ran and developed AB tests to showcase Network Next’s latency reducing technology and its benefits● Completed various requests for game studios, such as game launch support analytics

● Automated various company alerts using Airflow, Snowflake, Redshift, and Slack● Optimized N-offer sequence revenue using analytical, regression, classification methods. Project required weekly testing and reconfiguration of offers to send to product team to test. ● Performed regression modeling for user selection recommendations● Developed a BG-NBD probabilistic model for customer lifetime value that showed promising results on a small 100,000 test population with 90% accuracy● Completed ad hoc requests for multiple dashboard, chart, and analysis work for departments and people

● Write machine learning notebooks and scripts in Python for exploration and deployment● Write SQL and data models to perform analyses and other data manipulation tasks● Project 1: Creating a visual and interactive analytic dashboard to allow my clients to gain insights to make business decisions.● Project 2: Looking at a public corporation's customer purchase data. The company is well known in the clothing/fashion industry. The goal was to leverage purchases to inform future marketing decisions.● Project 3: Using machine learning regressors and classifiers to solve multiple issues, like identifying features related to binary and multi classification problems and multivariable regression analysis. ● Project 4: Ad-Hoc classification problem for big networking clientMuch more feel free to contact me to learn more about specific projects.

Completed a project involving multiple iterations of data munging, feature generation, and machine learning for Komodo Technologies

Worked with client in developing a classification model using 10 years worth of NFL game datasets. Project is in the beginning stages. Assisted in writing solutions and grading assignments and exams for a Multivariate Statistics and a Data Science Course. The topics include, but not limited to, Principal Component Analysis, K-Means Clustering, Hierarchical Clustering, and more.Analyzed car rental data for client to see what features were used more for rentals using PCA, K-means clustering, and bayesian statistics.Analyzed client’s Google Analytics data in order to understand user behavior. I applied regression analysis, clustering methods, and basic data analysis on the dataset.

Phoenix, Az, Us

Identified(top 4% among 3000+ prospective data scientists) for skills in statistics, mathematics, data analysis, and computer science to participate in a rigorous 8 week data science fellowship program. Throughout the program, I developed my project that uses Riot Games' API in order to analyze, aggregrate 5.4 GB of match data. I used the in-game features in order to predict the win percentage of a team during the course of a match. Each week we completed miniprojects related to tools used in the industry: 1. SQL: Aggregated and analyzed relational databases containing 530,000 NYC restaurant inspections data from Yelp to extract statistics on inspection grades and violations across various locations and cuisines, using SQL2. Machine Learning: Predicted a venue's popularity from information upon a venue's opening with machine learning. The 34 MB dataset comes from Yelp. The star rating was predicted with city, category, and latitude and longitude. The models were built using transformers and regressors. 3. MapReduce and Hadoop: Analyzed character entropy of extracted words and n-gram statistics of Simple English (320 MB) and Thai (900 MB) from Wikipedia, using mrjob, MapReduce, and Hadoop4. Natural Language Processing: Developed regression model using machine learning to predict Yelp ratings based on metadata analysis on 38,000 venues and processed text analysis on 1,000,000 Yelp reviews5. Time Series Analysis: Developed time series analysis to predict temperature in major US cities based on Fourier analysis of historical weather data over 525,000 time points in 12 years6. Spark: Using 10 GB of posts, users, and votes data from Stackoverflow to answer statistics as well as predict tags of questions based from body text using a logistic regression model.

Berkeley, Ca, Us

Majorana and KATRIN Experiments (Nuclear and Particle Physics Research)• Distinguished noise in MAJORANA Low Mass Front End Electronics wrt to experimental parameters. Best performance occurs with a minimal drain to source voltage of 0 V• Ran Monte Carlo simulations and calculated and analyzed multiple 10 gb datasets to produce probabilities due to various systematic effects to investigate non-adiabatic effects in the KATRIN spectrometer• Determined that KATRIN needs to increase the magnetic field strength from 3 G to 15 G in order to achieve 100% transmission rates for sterile neutrino searches. Thesis link can be found here: http://sfsu-dspace.calstate.edu/handle/10211.3/146110

San Francisco, Ca, Us

-Taught and lectured General Lab Physics focused toward non-calculus classical mechanics, electricity and magnetism, thermodynamics, and optical physics-Responsible for grading lab reports and creating weekly physics quizzes

West Lafayette, In, Us

Dark Matter Research (Particle Physics and Cosmology)• Investigated academic inconsistencies between peculiar velocities of the Sun and Milky Way for dark matter particle annual modulation research• Calculated yearly WIMP particle velocities and error for four different academic results. Used parametric regression analysis to find curve fittings. Results showed maximum at December 1st and minimum at June 2nd

West Lafayette, In, Us

High Energy Astrophysics Research• Ran control tests and soldered photomultiplier tubes for the Very Energetic Radiation Imaging Telescope Array System telescope. Constructed energy spectrums with (<1gb) datasets with error propagation and polynomial curve fitting• Ran Monte Carlo simulations for photons and calculated the χ2 values using Bessel’s corrections

Los Alamos, Nm, Us

Dark Energy Research (Cosmology)• Studied the efficacy of different parameters in discriminating theoretical models of dark energy using current and simulated datasets of 10-20gb.• The geometric parameters give the most accurate information about the nature of dark energy by two orders of magnitude less than physical parameters. Complete results can be found here: http://arxiv.org/abs/1012.1591

Master Of Science (M.S.), Physics

Bachelor Of Science, Applied Physics In Computer Science And Astrophysics