Sangho Lee
Biography
2021 Jul - : Computer vision research & development
2020 Dec - 2021 Jul: Machine learning engineer at 11street
2019 Nov - 2020 Nov: AI engineer at Carrot Insurance
2013 Jul - 2019 Oct: Data analyst at Samsung Electronics
MS of data science at Seoul National University
BSc at the Korea University (computer science and mathematics)
Research Interest
- Video representation learning
- Generative modeling
Highlights
- Proficient programming skills in Python, C++, R and SQL
- Experience with building advanced statistical methods including neural networks predictions in a big data environment
- Experience with dealing both structured and unstructured data (Image, video, text, etc.)
- Comprehensive knowledge of Calculus, Linear algebra, Probability, Computational algorithms and Dynamic programming
- Self-motivated and consistently strive to improve performance
Skills and Certification
- Programming: Advanced in Python with Pytorch, Tensorflow, Keras and Scikit-Learn, Intermediate in C++ in Linux/Windows
- Mathematics: Statistics, Linear algebra, Numerical analysis, Differential equation
- Data: Statistical understanding of structured/unstructured data (Image, video, text, etc.), preprocessing, analyzing and modeling
- Paper Implementation: Review and Implement ML/Deep learning papers (github.com/puhuk)
- Scholarship: National Science and Engineering Scholarship (Mar.2006~ )
- Others:
- OpenUP (Korea national opensource support center) frontier developer (Jun.2021 ~ )
- Deep learning project mentoring (AI bootcamp by Ministry of Employment and Labor w/ Samsung Multicampus & SK) (Nov.2019 ~ )
- Google machine learning bootcamp (Jan.2021)
- Tensorflow Developer Certificate (Oct.2020)
- Sales forecast competition contest top 10% model (Samsung Electronics) (Sep.2019)
Work Experience
Dec.2020 - 2021.Jul
Machine learning engineer @11 Street (E-commerce)
Seoul, Korea
- Role: Improve search model
Nov.2019 ~ 2020.Dec
AI engineer @Carrot insurace (Insurance)
Seoul, Korea
- Role: Insurance product development based on deep-learning technology
- Project
- Develop crack detection AI model for mobile phone insurance (Dec.2019~)
- AI detection model to detect crack on customer’s mobile device
- Extracts frames from videos and classifies screens, detects crack on the devices with CNN models
(http://www.joseilbo.com/news/htmls/2020/08/20200803403275.html)
- AI module implementation for classifying dog breed in Pet insurance (Jun.2020~)
- Develop dog breed classification model
- Develop crack detection AI model for mobile phone insurance (Dec.2019~)
Jul.2016 ~ Oct.2019
Data analyst @Samsung Electronics AI team (Manufacturer)
Suwon, Korea
- Role: Statistical analysis of user data and usage prediction (Samsung Pay, Bixby)
- Project
- Bixby user utterance analysis and market response prediction
- Analyze usage data of Bixby from European market with natural language processing algorithms - Prioritize services / languages in each market for increasing retention rate and number of active users - 15% increase in Monthly Active Users has achieved
- Service promotion strategy planning
- Market clustering based on usage data and prioritize marketing target and planning for global expansion - Analyze correlation between each factor from usage/market data
- Bixby / Samsung Pay business model setup
- Predict user and amount of each service per countries based on historical data (Device expansion, churn rate, etc.)
- Bixby user utterance analysis and market response prediction
Jul.2013 ~ Jun.2016
Data analyst @Samsung Electronics Procurement team (Manufacturer)
Suwon, Korea
- Role: Semiconductor demand forecasting
- Project:
- Demand prediction based on product lifecycle, market response, sales history and seasonality
- KNN clustering for market/semiconductor segmentation, Linear regression for long-term prediction
- Prioritize semiconductors supply chain by with device/semiconductor correlation analysis
- Stock amount decreased (over 10%) and Just-In-Time score has increased
- Demand prediction based on product lifecycle, market response, sales history and seasonality