zeki -

● Solved the bugs that created a conflict in the Microservice using FastApi. ● Restructured the DataBase logic. ● Integrated the Pytorch AI model solution (Instance Segmentation) into the microservice. ● Successfully communicated the asynchronous multiple requests using the queue system via RabbitMQ

Worked on Algorithmic trade on crypto platforms such as Binance, Deribit. ● Created the trading algorithms for the exchange strategy. ● Monitored the endpoints with FastAPI. ● Managed the application's backend and database. ● Used the pre-commit files for each project. ● Styled with pep 8, flake8, black in pre-commit pipeline. ● Tested Unit, Mock, Integration Test cases in Pytest. ● GitHub action was used to implement CI/CD pipeline. ● To keep requests in secure, created Cors proxy server in Node.js

Instabase Platform and its unique Python syntax and libraries have been learnt to use the Data Engineering Process. ● In order to understand the customer's requirements detailly, it was discussed by the customer to define the business logic. ● Using the state of art OCR (Optical Character Recognition) AI model, documents which have an extension such as pdf, jpeg etc were processed, then extracted all text data from the document. ● After considering which token is necessary for the customer from unstructured data. From the document, Unstructured data was converted to structured tabular data using OCR model and customized Python UDF function based on customers needs. ● Once all the engineering stuff has been completed, unstructured documents such as pdf, jpeg etc. were delivered to customers as curated and structured table data.

Made a contribution to developed a Spacy Rule-based Named Entity Recognition System with social science text data by using OOP (Object Oriented Programming). ● Saved annotation into Postgresql, by using Python configuration with Postgresql. ● In order to speed up code performance, Python Multithreading was used. ● The results were evaluated with a Subject Matter Expert to confirm the correctness of the system. ● Implemented attention based Hierarchical Text Classification with tensorflow. ● As a data science team, researched open-source data from a variety of resources to feed the system with the new data. ● Provided GPU support to Deep Learning Model (BERT) for the text classification. That significantly reduced model training time.

Built an automated data pipeline on AWS . ● While web scraping, Pandas, BeautifulSoup were used. ● Connected different AWS services to each other. For instance, EC2 to S3. ● In order to create an algorithm continuously, use Bash Script. ● Performed EDA on real estate data-set. ● Followed the kanban project management method.