Data Scientist
Afiniti
مجموع سنوات الخبرة :9 years, 9 أشهر
● Driving completion of deliverables and analytic deliverables
● Ensuring quality and reliability of data obtained from client sources and offline data sources
● Statistical testing and bias analysis with SQL and R programming
● Reconciling data (internally and with the client) and procedure with internal AI team
● Defining data joining procedures (collaborating with and audit procedures from AI team) for new installations and expansions. Document and get client to agree to procedure.
● Confirming that AI team is able to optimize the performance metric desired by the client and that the data provided by the client is sufficient (completeness and usability)
● Defining data validation reports to be shared with client
● Showing client how to look at the data and give them comfort around the procedure and the various checks that are part of validation. (Presenting results and SQL procedures to client via web session)
● Working as an analytics consultant for MasterCard and supporting Retail Credit Card Sales department of
one of the largest banks in Turkey.
● Developed SAS and Oracle scripts to analyze credit card sales by region, branch and channel on daily and
monthly bases.
● Collaborated with Analytics and CRM Team to determine short-listed model parameters and developed a
Supplement Card propensity model to enhance sales campaigns. Used SAS Enterprise Guide for data
extraction, preparation and SAS Enterprise Miner to determine variable features.
● Created several analytics dashboards with QlikSense and SAP Business Objects to monitor credit card and
debit card sales and stock.
● Performed detailed credit sales channel churn analysis by using Advanced Query Tool and SAS Enterprise
Guide to determine which channels underperform.
Trendyol Group (largest and fastest growing mobile commerce company in
● Worked on the development of a wide range of propensity models to predict customers category/sub-category based online-purchasing behaviour and email churn, and to personalize email campaigns.
● Developed adaptive propensity models for the Child & Baby segment using the purchase history of customers and the estimated age/month of the baby/child.
● Crawled browsing history logs to retrieve the most searched products information and developed recommendation and text completion models to recommend customers new products.
● Implemented deep learning algorithms with Keras, Tensorflow and OpenCV and reduced the workload of content feature entry teams by building CNN models to automatically detect the color and position of the product.
● Performed Skin Detection and Background Subtraction techniques to detect human figures in product images and eliminated product color misclassifications.
● DevOps experience with IT and Engineering teams while implementing predictive models
● Modeled risk parameters of Credit Conversion Factor (CCF) and Exposure at Default (EAD).
● Implemented IFRS9 and IRB compliant risk parameters of CCF and EAD for non-cash loan, overdraft
and credit card portfolios.
● Periodically assessed and improved risk management approaches requested by regulators (BRSA &
Basel) and consistently achieved model improvement percentages of 1-2%.
● Worked together with the DevOps teams and pioneered the technical integration process of
behavioral scorecards into a new software application.
● Designed and implemented Proof of Concepts for Telco and Banking sector customers by using Text Analytics including SAS Text Miner, SAS Sentiment Analysis and SAS Contextual Analysis to help Account Executive team to sell the related product
● Implemented survival analysis to inspect therapy effects of Regorafenib in patients with colorectal cancer for one of the largest pharmaceutical companies. Survival analysis was performed by using SAS procedures.
● Developed and scheduled credit risk validation scripts to monitor risk parameter model outputs by using SAS Enterprise Guide and SAS Model Manager.
● Received training in wide range of SAS Solutions including SAS Enterprise Guide, SAS Enterprise Miner, SAS Text Miner, SAS Model Manager, SAS Visual Statistics and Analytics.
● Worked in both software development and Watson Content Analytics teams to increase the accuracy of
IBM Watson Content Analytics’ new language detector for Turkish.
● Created keyword-based data crawling scripts with Python using Scrapy and BeautifulSoup to crawl
Turkish news, blog posts, Facebook and Twitter to assemble a Turkish Corpus.
● Performed Lexical Analysis, Part of Speech Tagging (POS) on a Turkish corpus and used state of the art
sentiment analysis methods.
● Achieved accuracy levels of over 90% and 70% in language detection and sentiment analysis models
respectively, and deployed two SaaS products.
Bachelor's Degree with Honors
Bachelor's Degree with Honors