Peut être disponible
(Mis à jour 2021-12-27)Senior Data Scientist
İstanbul, Türkiye
Natif English
- + 10 years of experience in Python
- + 10 years of experience in Machine Learning
- + 10 years of experience in SQL
Compétences (21)
Data Management
C++
Data Compression
DATA SCIENCE
ROBOTICS
CLUSTERING
FAULT DETECTION
C
Natural Language Processing
MECHANICAL DESIGN
Big Data
Deep Learning
COMPUTER VISION
Data Analysis
.NET
SOFTWARE DEVELOPMENT
Artificial Intelligence
Machine Learning
SOLIDWORKS
NLP
SENSORS
Résumé
My Main title is Senior Data Scientist with focus on developing R&D projects based on the Machine Learning area. Additionally, I have experience in academic research, robotics and teaching which gives a multi direction perspective for projects. Some of my focused areas are deep learning and gradient boosting machines in the machine learning topic. Wide variety of works are completed over a decade such as; artificial intelligence supported robots, academic papers, computer vision projects, NLP projects, classification of sensory data, time series analysis/prediction and fraud detection (outlier detection). Thus, these projects include image, text, columnar data and sensory data. I have a very unique background which gives a unique perspective for data science projects including feasibility, exploratory data analysis and model development steps.
Expérience professionnelle
2019-11 - Actuel
● NLP: I completed a project which consists of predicting the polarity of performance evaluation comments of employees. This dataset has a very low number of entries.
Thus I implement data augmentation, oversampling, embedding methods (deep learning) and some classification methods. Another NLP project is to find relevant tickets of IT support requests. Since this is a similarity problem, I applied some similarity metrics along with embedding/vectorizer methods.
● Prediction: I developed a model to predict required spare parts with the information gathered in early phases when the trouble is directed to the technical service. In this project some classification methods are used.
● Time series: I did exploratory data analysis and a fault detection model on sensory data comes from production lines. I implement some classification methods after a lot of data processing steps.
2017-08 - 2019-11
● Bionic Prosthesis: I made mechanical design and software development in tih TÜBİTAK supported project. Mechanical design was done by Solidworks. Robotic system software and AI softwares implemented in python. There are two AI software in the system; command detection from EMG (muscle) signals and grasp detection from objects (vision). AI modules uses deep learning methods as classification and prediction methods. At the end of the project, most dynamic prosthetic hand/arm was developed with the help of AI software.
● Smart Recommendation System: A service-based recommendation system was developed in order to serve e-commerce websites. Main development environment was python and PostgreSql. Additionally, google cloud systems, linux shell commands and DNS server were used too. In this project a framework is developed in order to dynamically manage the system (new server installation, new website, ).
Data management, system software and recommendation system was developed in python.
2019-01 - 2019-07
● Türk Telecom Fraud Detection Project: This project is aimed to detect frauds done by TT dealers, customers and personnels. In this project the main focus is preventing fraud with the help of big data and AI technologies. In this project, I worked in the R & D, analysis phase. I also develop machine learning models and develop a framework in order to manage and execute multiple models. Development language was Python.
Developed models are statistical methods, clustering, anomaly detection and deep learning.
2014-11 - 2016-03
TARBİL is a Turkish Ministry of Agriculture and Istanbul Technical University collective project to supply software / hardware needs for farmer support systems. The TARBİL project includes farm machine softwares, tablet pc softwares for farmers, central data collection / filtering / evaluation softwares and all related services. As its core purpose, TARBİL collects data from the field, evaluates it as a central authority and provides smart, automated support to farmers and related government units. I work with a small team in TARBİL to develop data evaluation and filtering processes, implementation of academic methods to real data, Prediction of agricultural events (Meteorological events, Pest risks, crop diseases, water needs,). In these areas, my work includes analysis, design, implementation and performance optimization.
● Interpolation Algorithm Development: Interpolation algorithms on location based sensor data were enhanced. These interpolation algorithms were used to create maps from location based sensors. Development was done with C # and .NET technologies. @@
● Data Filtering: 6 phased data filtering was applied to the sensor Datastream in order to filter out erroneous readings. These works include fine tuning of parameters too.
Development is done with Python.
2010-01 - 2011-01
2011-01 - 2011-01
Parcours scolaire
2016-01 - 2024-09
2011-01 - 2014-01
2006-01 - 2011-01