Elcin Ergin, Ph.D.
Data Scientist
Data Science Experience
Jan 2023-Current Chief Data Scientist - ORO Health, Montréal
Developing deep learning algorithms for skin disease detection and triaging for moles
Integrating in-house AI frameworks with LLMs to create conversational dermatology assessment and recommendation tool
Working on the productization of AI frameworks
Communicating between medical doctors and software development team
Overseeing data science projects
Sep 2020-Dec 2022 Data Scientist - ORO Health, Montréal
Developing deep learning algorithms for skin disease detection
Achieving over 90% accuracy in top-3 prediction with 42 skin diseases
Working on the productization of AI frameworks
Jan 2020-Sep 2020 Junior Data Scientist - AuDela, Montréal
Developed and supported classification and clustering based machine learning algorithms on telecommunication sector
Supported the productization of AI frameworks
2014-2020 PhD Researcher - McGill University, Montréal
Investigated the location-specific inventory and sales dynamics by applying econometric models and machine learning techniques on large data sets, resulting 2% increase in revenue for a fast-fashion retailer located in Europe
Published an A level journal article
Using color histograms on image data, constructed a novel algorithm to optimize color assortment scheme for online stores
Gave lectures on operations management
2012-2014 MSc Researcher - Middle East Technical University, Ankara
Developed parametric, nonparametric and hybrid approaches to improve short term electricity forecasting, achieving a 17% more accurate prediction of hourly electricity demand for an energy solutions company in Turkey
2011-2012 Cash Management Optimization Project - IsBank, Istanbul
Optimized the cash management system of a bank operating in Turkey as the final year undergraduate project
As a team, optimized the number of central cash offices and their locations to minimize the total operational cost (by 18%) and to improve the utilization of the system (up to 50%) by modeling it as set-covering and travelling salesman problems and approximating using vehicle routing algorithm