Idrak Asadbayli
Data Scientist | Machine Learning | Deep Learning
Contact MeProjects
Pre-Training and Fine tuning
Date: December 2024
Collecting unique sentences in the job posting field with web scraping. Building a deep learning model using the pre-training feature of the Bert and T5 algorithms. Building a specialized LLM model by fine tuning the BERT and T5 algorithms with a specific dataset. Since these steps use a lot of resources, use the GPU and TPU capabilities of Kaggle and Colab.
Passing Score Prediction
Date: November 2024
Predicting university entrance scores based on past years, specific characteristics of universities, and scores obtained by students. Providing appropriate advice to students who want to make choices. This system is based on the GRU deep learning model, taking into account the capabilities of time series and neural networks.
Sentence Classification
Date: September 2024
Sentences are classified by deep learning modules. For this, RNN, GRU and LSTM models are used together. The results are combined through Ensemble Learning to try to achieve the best result.
Smart Irrigation System
Date: May 2024
It is a direct and remote controlled irrigation system. This device controls the motor and valves to ensure the irrigation of the garden. The central unit can be controlled from a phone or a monitor on it, and an alarm system can be set up so that it can irrigate in the future. This not only reduces human work, but also helps protect nature by reducing water consumption in irrigation.
Face Recognition
Date: July 2023
It consists of 3 sections: - Introduction of images to the system - Recognition of the person in the image - Recognition of the person in the video
ETL from JSON to PosgreSQL
Date: October 2022
I organized and cleaned the data within the JSON file, ensuring it was properly structured and ready for use. Afterward, I loaded the data into a PostgreSQL database, breaking it down into multiple tables to create a relational schema. Relationships were established between the tables to maintain data integrity and allow for efficient querying within the database.
ETL from CSV to MySQL
Date: March 2022
I processed the data from the CSV file by cleaning it and removing any duplicate entries. I divided the table into multiple smaller tables and established relationships between them to ensure a well-structured and normalized database design. After ensuring the dataset was properly prepared, I successfully uploaded it to a pre-configured MySQL database.
RFI creator
Date: Septenber 2021
The table containing the RFI data, the address where the RFI will be created and the RFI numbers are selected. It is possible to create more than one RFI at the same time. RFI numbers can be recorded sequentially or mixed.
Biometric door system
Date: November 2018
Checks the accuracy of the employee with a fingerprint sensor. If the employee's fingerprint is in the system, the door is opened and the employee's start time is sent wirelessly to the central system. This wireless connection is made by the HC12 module.
Online control system
Date: October 2018
Connects to the internet via wifi or ethernet port. Manages device-related equipment through a website. A simple example: turn the LED on and off.
WakeMeUP
Date: Septenber 2017
The eyeglasses constantly monitor the worker and activate an alarm to wake them up if they fall asleep. In addition, it provides tracking of the movement of vehicles for logistics companies.
