S. Başak Anayurt Kobrin
Data Scientist at Grabango- Claim this Profile
Click to upgrade to our gold package
for the full feature experience.
-
English Full professional proficiency
-
Spanish Limited working proficiency
-
Turkish Native or bilingual proficiency
Topline Score
Bio
Credentials
-
Convolutional Neural Networks
CourseraSep, 2020- Nov, 2024 -
Improving Deep Neural Networks: Hyperparameter tuning, Regularization and Optimization
CourseraAug, 2020- Nov, 2024 -
Structuring Machine Learning Projects
CourseraAug, 2020- Nov, 2024 -
Neural Networks and Deep Learning
CourseraJul, 2020- Nov, 2024
Experience
-
Grabango
-
United States
-
Software Development
-
1 - 100 Employee
-
Data Scientist
-
Aug 2022 - Present
-
-
-
Alt
-
United States
-
Financial Services
-
1 - 100 Employee
-
Data Scientist
-
Jan 2021 - Feb 2022
-
-
-
Insight Data Science
-
United States
-
Higher Education
-
1 - 100 Employee
-
AI Fellow
-
Sep 2020 - Dec 2020
- Built a toxic content - spam, hate speech, self harm - filter for Commaful (a storytelling platform). - Reached to a F1 score of ~85% by utilizing convolutional NN and BERT for different tasks. - Served it as a Streamlit web app deployed in AWS with Docker - Built a toxic content - spam, hate speech, self harm - filter for Commaful (a storytelling platform). - Reached to a F1 score of ~85% by utilizing convolutional NN and BERT for different tasks. - Served it as a Streamlit web app deployed in AWS with Docker
-
-
-
Eurecat - Technology Centre of Catalonia
-
Spain
-
Research Services
-
500 - 600 Employee
-
Research Data Scientist
-
Dec 2017 - Apr 2019
• Developed analytical part of a data driven integrated tool for monitoring the efficiency of mine excavators. • Analyzed noisy, multivariate time-series data from a strain sensor placed on an operating mine excavator. • Developed an automatic algorithm for detecting events (e.g. removing soil and rocks from the surface) and eliminating spurious events due to sensor noise; key methods included wavelet transforms and gaussian mixture models. Deployed this algorithm as the core backend of a graphical, web-based user application. • Conducted exploratory data analysis on detected events to classify terrain properties and bucket motions. For this purpose implemented and tuned several machine learning models (e.g. random forest, SVM, Naive Bayes Classifier). Reached ~90% accuracy by generating effective custom features from multivariate time series data as well as implementing appropriate dimensionality reduction on them. • Consulted the client on designing and implementing new on-site experiments. • Supported mySQL-python interaction and predictive modelling of a plastic injection quality control system. Show less
-
-
-
Universitat Politècnica de Catalunya
-
Spain
-
Research Services
-
700 & Above Employee
-
PHD Candidate (Left ABD)
-
Nov 2015 - Oct 2017
Developed strategies for reducing the time cost of fatigue analysis of wind turbine components. Developed strategies for reducing the time cost of fatigue analysis of wind turbine components.
-
-
-
GE
-
United States
-
Industrial Machinery Manufacturing
-
700 & Above Employee
-
Research Collaborator
-
Nov 2015 - Sep 2016
Worked under an industry-academy collaboration during the first year of the PhD studies. • Conceptualized and applied a strategy that decreased the time cost of fatigue analysis of wind turbine structures up to 50%. • Developed a clustering algorithm to safely disregard the less damaging load cases (multivariate time series). This reduced the load cases to be processed in fatigue calculations up to a factor of 2. • Researched on representing each load cluster by a single load case that would preserve total fatigue features. For this purpose, identified time series models (e.g. FS-TARMA, Gaussian Process, Markov Model) that can reproduce such fatigue load cases. Show less
-
-
-
Alstom Power
-
Electric Power Generation
-
700 & Above Employee
-
Research Collaborator
-
Oct 2014 - Sep 2015
• Delivered a data driven solution for the engineering team to assess if the parameters of real operating wind-turbines are as expected from the structural design. Developed a semi-automatic tool for identification of local structural dynamics and modal analysis with the use of sensor data from a wind turbine prototype. • Implemented and tuned parametric models (e.g. ARMAX, Box-Jenkins), stability diagrams and unsupervised learning techniques (e.g. Gaussian mixture models, spectral clustering) for identifying the structural black-box system and it's parameters. • Applied Kalman filtering to test the time dependency of the structural parameters. • Designed frequency response functions (FRF) that approximate real structures and simulated loads from them. These loads were used as test data for further validation. The tool reached an accuracy of ~90% in both approximated cases and real cases. • The tool worked with over 90% accuracy and it brought vital insight about the validity of assumptions of the structural design phase. • Published “ Local structural dynamics identification in offshore wind turbines based on experimental data” , EWEA Annual Conference, November 2015, Paris - http://toc.proceedings.com/29663webtoc.pdf ) Show less
-
-
-
Leitat Technological Center
-
Spain
-
Research Services
-
300 - 400 Employee
-
Intern Engineer In International Project Office-Transport & Aeronautics
-
Jul 2014 - Sep 2014
-
-
-
FNSS Savunma Sistemleri A.Ş.
-
Türkiye
-
Defense and Space Manufacturing
-
700 & Above Employee
-
Design Engineer
-
Mar 2013 - Aug 2013
-
-
Project Assistant
-
Apr 2012 - Mar 2013
-
-
-
Ford Otosan
-
Türkiye
-
Motor Vehicle Manufacturing
-
700 & Above Employee
-
Intern Engineer
-
Jun 2010 - Jul 2010
-
-
-
Herrenknecht AG
-
Germany
-
Machinery Manufacturing
-
700 & Above Employee
-
Intern Engineer
-
Aug 2009 - Sep 2009
-
-
Education
-
Universitat Politècnica de Catalunya
Doctor of Philosophy (PhD), Industrial and Control Engineering -
Universitat Politècnica de Catalunya
Master of Science (MS), Aerospace Enigneering -
Orta Doğu Teknik Üniversitesi / Middle East Technical University
Bachelor of Science (BS), Mechanical Engineering -
Ankara Atatürk Anadolu Lisesi
-
TED Ankara Koleji