Ece Aktan Hatipoglu

AI Research Engineer, based in Trabzon, Turkey

Email / LinkedIn / GitHub / Medium / LeetCode

Experience 👩🏻‍💻

Senior Data Scientist @ Hepsiburada (October 2022- present)

• Contributed developing Hepsiburada’s AI models, enhancing search and findability, focusing on query classification and product ranking.
• Developed, and maintained an end-to-end machine learning pipeline that utilizes NLP techniques for keyword based fast review generation. The pipeline includes Airflow for serving and has significantly improved both the quality and quantity of reviews.
  
  

Reseaerch Engineer @ Huawei (September 2019- October 2022)

• Developed a multilingual spelling correction pipeline for the search module of Huawei App Gallery, including domain adaptation of BERT, keyword extraction for app visibility and search terms ranking, serving and version management through Huawei modelarts .
• Implemented a parallel corpus filtering pipeline in collaboration with AARC of Huawei, resulting in a published ACL paper and a ranking in the EMNLP contest of WMT20.
• Technologies used: Python3.x, TensorFlow, PyTorch, NLTK, SpaCy, Scikit-learn, Pandas, Jupyter.
• Models used: Symspell, DeepPavlov, SimCLR, Moco, Contrastive Learning, Yake, Bert, m-Bart.
  
  

Data Scientist @ Getir (May 2018- July 2019)
Getir is an online retails delivery app which promises less than an hour delivery. Their business model is similar to Glovo, only Getir has its own stores in each crucial spots in Istanbul where they keep/manage wide-definite products. The demand prediction and planning, couriers planning & assignments, marketing analysis are all carried by the Data team.

• Solved one of Getir’s major problems of predicting store based daily orders using Tensorflow’s LSTM implementation, achieving a 2% error rate(weekdays).
• Employed a regional hexagonal gridding system to carry out location-based analyses and store region management.
• Conducted campaign effect and precision clustering, geographic rival and risk factors analyses, and ad-hoc analyses to observe client RFM, campaign success, and store region productivity.
• Implemented a sentiment analysis model to track public reliability of the company by mining tweets that mention the company using Twitter dev API.
• Technologies used: Python 3.x, TensorFlow1.x, AWS, S3 buckets, MongoDB, RedShift, SQL, Jupyter Notebooks, Shapely, Tweepy, Scikit-Learn, Statsmodels.
• Models used: Bayes, ARIMA, SARIMA, LSTM, Clustering.
  
  

AI Research Engineer @ Etiya (Dec 2017- May 2018)
Explanation

• Developed a text-CNN model to identify street language and mocking in Turkish, achieving an 88% accuracy.
• Mined task-specific data using Python’s Scrapy and Tweepy.
• Technologies used: NLTK, SpaCy, Python3.x, TensorFlow1.x.
• Models used: TextCNN, Word2vec.
  
  

Junior PHP Developer @ Ubit (Jul 2014- Sept 2015)
Worked as a backend PHP developer, for their school management platform called STOYS

• Designed the back-end of the web pages for STOYS, a school management platform.
• Contributed to the redesign of database tables and relationships during a version upgrade of the STOYS system, ensuring efficient data management and improved system performance.
• Technologies used: PHP, Zend Framework, SQL.
  
  

Research Assistant @ BAU Computer Vision Lab (Jul 2015- Sept 2017)
I participated in the BAUFera project, a university research initiative funded by TUBITAK. The project’s objective was to develop an emotion detection system from faces in surveillance cameras. As a team member I,

• Implemented a multi-task deep learning architecture for gender classification and facial landmark localization.
• Technologies used: TensorFlow, Matlab, Python.
  
  

Education 📚

Computer Engineering
Bahcesehir University - Istanbul, Turkey (2018-Droped out at thesis)

• Major courses completed: Data mining, Computer vision, Machine learning.

BS in Software Engineering
Bahcesehir University - Istanbul, Turkey (2010-2015)

Site Projects 🐝

• 2022 Finding the trace of adverse childhood experiences in reddit micro blogs.
• 2020 Accident prediction on notifications posted by local municipality on Twitter.
• 2018 Advertising placement for the Twitch game streams.
  

PUBLICATIONS 🔦

• [1] H.Acarcicek, T.Çolakoglu, P.E.A.Hatipoglu, C.H.Huang,and W.Peng. Filtering noisy parallel corpus using transformers with proxy task learning. In Proceedings of the Fifth Conference on Machine Translation, pages 940–946, 2020.
• [2] P.E.Aktan, G.Hatipoglu, and N.Arica.Risk classification for breast cancer diagnosis using her2 testing. In 2016 24th Signal Processing and Communication Application Conference (SIU), pages 2133–2136. IEEE, 2016.
  

Note 🖌

Special thanks to Carolyn for sharing the knowledge.