Welcome! I am Genesis 🐙 and I'm a recent Duke graduate of a master's degree in Interdisciplinary Data Science. I'm interested in data science and its intersections with public policy, healthcare, and journalism. Stay and learn a bit about my story and enjoy some of the projects that I have worked on.
Here are some snippets of my professional and personal story 🦦. I grew up in Nanjing, China and Portland, Oregon 🌲. I love using data and code to tell stories and make beautiful visualizations. But i'm not only passionate about data, I also love journalism, music, baking and reading. You can find my GitHub projects as well as some of my clips at The Bruin in the slider below.
I double majored in Statistics and Political Science at UCLA 🐻 . I enjoyed applying data science to analyze and catalyst social good.
At UCLA, I found a love for journalism, first as a news reporter and then a news editor and finally editor-in-chief, where I ran and won on a platform to improve diversity in the newsroom and its coverage.
I joined UCLA VRP to my knowledge in public policy and my skillset in data science to create some social good. I used these skills to help the center's incredible legal team defend voting rights in Washington, New York, and Kansas.
After spending two summers interning at H&R Block and learning more about ML/DevOps. I joined Duke to dig deeper into the nuances of MLOps, Deep Learning, and Data Enginnering, and to keep learning and growing as a data scientist.
I am a sucker for a good'ol historical fiction or historical analysis. My favorite book I read this year was Cloud Cuckoo Land by Anthony Doerr. You can follow along with my reading list and my book reviews. Feel free to email me book recs as well!
I'm pursuing full-time data science roles, excited to apply my repertoire of skills and experience to make an impact!
This full-stack database project is a Hunger Games-inspired shopping platform for tributes, modeled as a mini-Amazon. This interactive site allows tributes to purchase survival gear and supplies using virtual currency. I was responsible for integrating user accounts, product listings, and transactions, ensuring seamless functionality across the platform. The project involved collaboration on database design and code integration, enhancing the shopping experience under simulated high-stress conditions for users.
This study explores the possibility of reversing the relationship between text and image using advanced Vision-Language Pre-training (VLP) frameworks, such as the BLIP and ViT models. Our findings demonstrate that the fine-tuned BLIP model significantly surpasses the zero-shot baseline, showing a 25% improvement in average cosine similarity scores on the test set.
To better understand the common problem of motion artifacts in patients with breast cancer, we used OpenCV to simulate motion artifacts and a custom-built Faster R-CNN computer vision model to perform object identification and localization on tumor images to determine the direction of the motion blur that impacts classification performance the greatest.
This HackDuke project achieved third-place in the environmental track. In it, we used a state-of-the-art image-to-text neural network model, natural language processing, and data analysis to help uses identify the hidden electricity, water, and CO2 costs associated with a piece of clothing. Designed with sustainability and community in mind, the app allows users to conceptualize the environmental impact of their clothing purchases, compare them to the clothes of other users and make more informed decisions. The app is deployed on Heroku and is still in development.
This FastAPI application, developed and deployed through AWS Cloud9, Elastic Container Registry, CodeBuild, and App Runner, queries a database of goodreads books and provides its users with information on books' authors, publisher, rating, and number of reviews. Continuous Integration and Continuous Deployment are performed through Actions and CodeBuild, pushing the latest images to AWS ECR and App Runner. To save cost, the App Runner is on pause, and users can replicate the app by running the python script or the docker image on DockerHub.
In this project, I used Python's Tweepy in conjunction with SQLite to automate the extration of tweets on Joe Biden's timeline and conduct data summarization and exploration using SQL queries.
This project uses Dask parallelization and time-series data on the temperature of over 3,000 cities globally to estimate future temperature in these cities in order to understand the disproportionate impact of climate change globally. The project was developed using GitHub Action's Continuous-Integration/Continuous Delivery framework. Users can access model predictions through a Click command line app or a Streamlit web app. Both apps are containerized and can be run from my DockerHub remote repository.
Award-winning project in the UCLA Datafest competition where our team provided inference analysis for our client at the Rocky Mountain Poison and Drug Safety Center. Using a random forrest classifier and splitting the participant behaviors into corresponding risk levels, we found that recent use of cannabis is a major predictor for high drug overdose risks. The project won the Best Visualization award in the competition and achieved third place in International Statistical Institute Florence Nightingale Prize.
In this project, I aimed to understand whether presence of African American churches strengthen Black voter turnout in the South using county-level data from the Association of Religion Data Archives on the attendance and presence of predominantly African American churches.
Click on the slides below to read some of the stories I wrote and those that were written about me during my time at the Daily Bruin. As the first person of color to take charge of the newsroom in almost a decade, I was able to bring a new perspective to the newsroom and the stories we told. Photo credits to the incredible photographers at The Bruin, their names are in the stories.
Tracing back the reasons I joined the Daily Bruin, I must admit that ego was a part of it.
UCLA faculty criticize decision to let LAPD use Jackie Robinson Stadium to detain Black Lives Matter protestors (Photo Credit: Liz Ketchem)
For the third consecutive year, the Daily Bruin is reflecting on its progress toward becoming a more diverse and representative center of journalism for the UCLA community.
Bernie Sanders presents progressive platform to 5,200 supporters at East LA rally
Presidential candidate Klobuchar discusses domestic violence at UCLA event
The California secretary of state said his office is working on increasing voter participation and enforcing voter security at a university event Thursday.
American public schools are increasingly segregated along the lines of race and income, a study found.
Providing hubs for students experiencing housing-insecurity only short-term fix
Q&A: New Equity, Diversity and Inclusion Vice Chancellor Anna Spain Bradley's goals for her upcoming term
UC Board of Regents votes to increase nonresident tuition by $762 per year
UCLA faculty discuss what mail-in voting will look like for the 2020 general election
The assistant secretary of state for public affairs for the United States encouraged students to pursue a career in the state department at an event Friday.