Sangseok Joseph Lee
People Data Scientist
Biography
I gained my master's degree in data science at Duke University. I am currently working with Computational Culture Lab (based on Stanford University and UC Berkeley) as a research assistant, mainly focusing on how to measure organizational culture from empirical data. In 2020 summer, I worked as a Jr. data scientist in Summery, a tech startup in Silicon Valley, helping match their charitable projects with their employees’ personalities. Before coming to Duke, I worked as a people analyst in KEPCO (Korea Electric Power Corporation), #1 global utility company with 20,000 employees, for four years. My interest lies in building predictive models, deriving business insights from data (specifically orgnization or HR data) and data visualization.
A few fun facts: I published my first book when I was at 17; I used to hire data scientist as a recruiter and one day I thought "Why not I become a data scientist instead of recruiting them?" and here I am as a data scientist with HR background.
Interests
- Statistical Modeling
- Recommender system
- Natural Language Processing
- Data Visualization
- Business Analytics
Education
-
M.S in Interdisciplinary Data Science, 2021
Duke University, NC, U.S.
-
B.A in Business and Economics, 2014
Handong Global University, Republic of Korea
Technincal Skills
-
Python, R, SQL, HTML
-
Tableau, Hadoop, Power BI, Git, Docker, Scikit-learn, Keras
Experience
Research Assistant
Computational Culture Lab in Stanford University
- Developing a model to measure organizational culture by calculating the similarities of vision/mission statements in public articles and comments from Glassdoor data
Jr. Data Scientist
Summery
- Developed ‘Net Culture Score’, a score of organizational cultural alignment between organizational culture and individual values by combining behavioral science with machine learning and Natural Language Processing (NLP), which has become a key indicator in Summery’s official report
- Built automated report generation (Java) with dashboards (Tableau) to visualize key metrics
People Analyst Director
KEPCO (Korea Electirc Power Corporation)
- Achieved 30% reduction of labor costs through developing ‘Chatbot Recruiter’ using NLP
- Saved 10% budget by improving recruiting process based on predictive workforce planning
- Developed predictive modeling of employee turnover warnings and solutions for managers
- Conducted pre-post analysis of talent retention to measure the impact of headquarter relocation
People Analyst
KEPCO (Korea Electirc Power Corporation)
- Enabled data-driven decision making by providing business insights, integrated dashboards, data storytelling, advanced analytics and designing metrics packages and KPIs using SAP Success Factor
- Developed and communicated key insights and strategic recommendations to the executive leadership around workforce planning, retention, performance and organization culture
- Implemented the first HR data science team for strategic workforce planning and talent assessment
Business Analyst
SBA (Small Business Administration) / Headquarter
- Developed a unified financial DB system by minimizing unnecessary aggregation work in all states in U.S.
- Discovered potential factors that lead small businesses to succeed by analyzing past ten years dataset, which resulted in 20% reduction of the failure rate of small businesses funded by SBA
Projects
Duke Undergraduate Academic Trajectories and Course Recommender Systems
Developed a personalized course recommendation system to help facilitate course load planning, order, and selection for future students leading to academic success
Duke Undergraduate Academic Trajectories: Network Analysis and Course Recommender System
Using historical enrollment and graduation data, we propose several tools to facilitate major and course selection for undergraduate students at Duke University.
X-Ray Detection of COVID-19
proposed a deep learning model, ResNet50, to identify COVID-19 chest X-rays
Detecting Deepfakes
reproduced some state-of-the-art models (ResNet-18, Xception, InceptionResnet V1), assessed the performance of these models and identified their challenges and limitations
Detecting Solar PVs in Aerial Imagery
Developed a model to identify solar photovoltaic (PV) arrays in the aerial image through Convolution Neural Network
Lecture on People Analytics for undergraduates
As an alumnus of Handong Global University (HGU), I presented my lecture to undergraduate students about People Analytics.
Speed Dating Analysis
Used hierachical logistic regression model to find the main factors to lead people(male and female)'s decision in the speed dating.
The Effect of Job Training on Wages
Analyzed the effect of job training on disadvantaged workers in the US using regression model
Estimating the Impact of Opioid Control Policies
pre-post analysis and a difference-in-difference analysis when comparing opioid shipments and overdose deaths.(One report for Data Scientist and the other report for a policy maker)
The Effects of Social Distancing on COVID-19
Analyzed the effect of social distancing on the number of new COVID-19 cases per day and the mortality rate with three merged data from FIPS, Census, Safegraph Data
Text analysis for social distancing during Covid
Analyzed Twitter, Google Trend, Safegraph data to see the impact of social distancing given the date of Stay-at-Home order