Innovation Team

- Conducted data science research at the U.S. Air Force Operational Energy to optimize fuel logistics networks and build operational plans that address risk- We used a multiple regression model to train our data. For every flow object, the flow along each edge is an input variable and the score for the entire flow plan is the output variable. We chose to use a regression model because it works best with the data points we have (various numerical features that correspond to a numerical output).- Conducted Data Science and Machine learning research at the U.S. Air Force Operational Energy: Used a Multiple Regression Model with Exploratory Data Analysis, Neural Network, Seaborn Graphs and Tensorflow to optimize fuel logistics networks and built operational plans that address risk; Found the maximum score to a flow object in the dataset was roughly 76%

Amazon Grocery: F3 In-Store & Analytics Team- Filtered 120 million data points with 30 attributes of customer behaviors and purchase history from Amazon Go and Fresh “Just Walk Out” technology (checkout-free) using SQL and Excel- Created an ETL pipeline to extract and store metrics from the Amazon Grocery API with Redshift, AWS S3, and Athena - Created 2 automated Tableau Dashboards with weekly refreshing schedules; Built a Tableau Data Extract data pipeline in Cloud Server to ensure real-time monitoring

- Built and provided back-end data filtering and analytics for IT finance data pipeline/dashboard/website with Rest API and Python dashboard/with Rest API and Python (Django, Jinja 2, Pyecharts); developed operational metrics and guidelines

Supply Chain Serialization Solutions Team- Created and analyzed Tableau Dashboards on profit & loss, risk data (market size, shipping dates, packaging levels)- Produced SQL queries on Aginity WorkBench to retrieve supply chain reports from shipping sites across the enterprise - Generated Splunk Search Processing Language (SPL) queries, Statistical Reports and constructed Dashboards using XML- Cleaned and transformed datasets containing over 500,000 data entries and 20 attributes from the Global Supply team- Selected the critical features using Lasso and Stepwise methods in a cross-validation framework in Splunk - Implemented Splunk Machine Learning Toolkit with machine learning models KNN, Logistic, Linear, Random Forest, and picked a comprehensive and high accuracy model to update weekly global data analysis & predictions

- Launched a product updated with 3 new user interfaces; compared KPIs in sales and analyzed 1k users’ data- Collaborated with UI designers on language keyword updates; delivered near 50% productivity gains ~ $1M in value- Worked on a competitor analysis using Excel that detailed 30+ unique data points of 23 direct and in-direct competitors

- Held and facilitated office hours helping students with debugging lab, homework, and projects; tutored SQL, Feature Engineering; tools such as Pandas, sklearn, and Seaborn to analyze data; machine learning models and visualizations

- Implemented an algorithm to find representative words from a website/Microsoft Word document/PDF with Natural Language Processing (NLP) library gensim; Built a web scraping and parsing pipeline with library Beautiful Soup - Quantified text importance data using algorithms of Calculated Inverse Document Frequency (measures how important a term is) and Term Frequency (measures how frequently a term occurs in a document)- Built a Gradient Boosting Classifier with k-fold cross validation with an accuracy of 85%; Created Dropbox data pipeline

- Collected, translated, and compared environmental policy data from the EU, Chinese, US Environmental Protection Agency (EPA), World Health Organization on water, soil, air restoration and remediation chemical standards on Excel- Presented potential optimal data visualizations to the research team on a weekly basis using PowerPoint and Excel- Extracted key data from all experiments in the university database using Excel and Python on Jupyter Notebook.- Assisted to design systems relying on the ability of microbes and their constituents, both naturally occurring and induced, to support natural attenuation of the environment.

- Completed group final project: “A Comparison of Statistics of SDSS (Sloan Digital Sky Survey) Galaxies to a Randomly Generated Patch of the Universe,” with applied extensive uses of programming language Python and indicated to find the randomness of galaxy distributions. - Participated in Cluster 2: Reasoning About Luck: Probability, Statistics, and Their Uses in Science.- Collaborated with UCI professors and graduate students from the Department of Physics and Astronomy. Generated multiple random variance samples with predictive modeling in the density of galaxies (Monte Carlo method) from Sloan Digital Sky Survey and compared; Found the variance of the SDSS galaxy number density was 2.2 higher than that of a randomly generated set of galaxies; Suggested that the Universe is not random, tends to be cluster

- Successfully assembled, tested and launched Irvine01 (a nano-satellite) into Low Earth orbit in January 2016. - Collaborated with students from Irvine Unified School District (Beckman, Irvine, Northwood, Portola, University, and Woodbridge high schools). - Participated in professional panel reviews with experts such as JPL personnel. - Engaged in a variety of orbital maneuvers and experiments, including operating the CubeSat to position the antennae, solar panels, and camera for optimal operation.- The first program of its kind in West Coast.Irvine01 CubeSat Communications Team; Wrote Cubesat data transmission programs in C++ implementing ARSFTP (Amateur Radio Software File Transfer Protocol) for efficient file transfer with minimal data decayCollaborated with CalPoly SLO to engineer custom encryption and decryption software for data transmissionSet up Sidekiq Software Defined Radio on Linux to allow for code to be written and downloaded to the RadioAssisted in filing paperwork to obtain FCC licenses for radio teamPresented the satellite’s radio transmission link budget during NASA Design Review for mission approval