Kim Sha - Data Scientist

Masters student at

Cornell Tech

Dual MS in Applied Information Science, Information Systems - Urban Tech

Currently focused on

not breaking this website

Top languages

Python

PostgreSQL

JavaScript

Main tools

Pandas
Scikit-learn
Plotly (D3)

About Me

Most recently a Data Science Intern at Square (Block, Inc.), I'm currently pursuing an MS in Applied Information Science at Cornell. I graduated from Princeton in 2019 with a BSE in Chemical & Biological Engineering, and was a Senior Data Scientist at CKM Analytix prior to Cornell. This website serves both as a portfolio and sandbox for me to write about data science in the context of urban technologies and sustainability.

Writing

Exploring NYC 311, Part I — 01 January, 2023

Identification, visualization and analysis of an urban soft system.

Projects

Predicting Service Level Agreement (SLA) Violations for NYC 311

Classifying the likelihood that government agencies meet response time commitments for 311 complaints, a 'hotline' that users can dial to submit service requests for government agencies to address.

Status: live
Python
Plotly
Scikit-learn
XGBoost
SQL (SoQL)

Fine-Tuning a Vision Transformer for Detection and Classification of AI-Generated Images

Identifying whether an image is authentic (created by a human) or generated by one of a series of text-to-image AI generators (i.e., Stable Diffusion, Midjourney, and DALL-E) by fine-tuning a pre-trained vision transformer (Swin-Tiny) to tackle a multiclass classification problem.

Status: live
Python
PyTorch
Hugging Face

Event Planning App

Web application to create and track details of upcoming events and vacations.

Status: live
JS
Express
React
Tailwind

Serverless Scraping of Solar Irradiance Data

The NREL Measurement and Instrumentation Data Center (MIDC) tracks granular solar and meteorological trends down to the minute at a station in Golden, CO. This automated crawler scrapes the data into a Postgres instance by leveraging AWS resources.

Python
PostgreSQL
Scrapy
Serverless
AWS Lambda
AWS RDS

DFFNN Binary Classifier

Implemention of a generic deep feedforward neural network (DFFNN), or multi-layer perceptron (MLP), to tackle binary classification problems. Collated from a series of assignments from Andrew Ng's (deeplearning.ai ) Deep Learning Specialization.

Status: live
Python
Neural Networks

K-Means Clustering of Solar Irradiance Profiles (Undergraduate Thesis)

Modelling the performance of near-UV organic solar cells under meteorologically representative spectral irradiances. Data was queried from the NSRDB (National Solar Radiation Database) API maintained by NREL.