Atilla B. Senior Data Scientist

Experience

Lead Data Scientist

Jul 2022 - Present

Diabetic Retinopathy

Description: Medical project for classification and segmentation the fundus images for
diabetic retinopathy signs.
Responsibilities:

• Initialized AWS Sagemaker infrastructure for continuous training pipeline of classification neural network;
• Reduced data preprocessing step time from 10 minutes to 4;
• Deployed custom JupyterLab server on the machine with datasets, that speed up access to the data twice and ease the process of data transformation with remote jupyter notebooks;
• Implemented image classification pipeline that employs EfficientNet architecture.
• Developed fully automated labeling infrastructure to manage image segmentation tasks for a team of 5 data annotators;
• Introduced image segmentation data-encoding method that reduced 60% of data transferring across API.

Technologies: AWS (Sagemaker, EC2), Tensorflow, OpenCV, Pandas, JupyterLab, Docker, Postman

Optical drone navigation

Description: At this stage the main task of the project is to create a solution, which will perform autonomous navigation of UAVs, using CV. This complex project consists of visual navigation, object detection, identification, tracking and UAV steering tasks.
Responsibilities:

• Research and analysis of different deep learning based techniques for the task of Single Object Tracking (SOT/VOT).
• Development of object tracking module based on solutions with neural networks;
• Development of object detection and identification modules;
• Creation and implementation of UAV steering algorithms.

Technologies: Python, PyTorch, Tensorflow, OpenCV, OpenMMLab, Docker

Spermatozoa tracking

Responsibilities:

• Writing commercial proposal;
• Doing preliminary research on what is possible to do and approximate methods.

Technologies: Python; Scikit-learn; OpenCV; Scikit-image; TensorFlow; AWS (Sagemaker, EC2, S3, Lambda, API Gateway, DynamoDB, RDS, EKS); methods: morphological operations, image binarization, classification, neural networks.

Hacking the Human Body

Description: Train a segmentation model on tissue samples obtained with certain methodology which would perform well on tissue samples obtained with a different methodology
Responsibilities:

• Researching existing methods;
• training a neural networks for the task.

Technologies: Python; PyTorch; TensorFlow; AWS (Sagemaker, EC2, S3, Lambda, API Gateway, DynamoDB, RDS, EKS); methods: neural networks (transformers, unet).

Room Flooring Design

Description: Develop a system which would replace the floor in an image with a given floor
Responsibilities:

• Design the pipeline of the system;
• develop a floor detection neural network;
• set the tasks and supervise the main team working on the project.

Technologies: AWS; Python; PyTorch; OpenCV; AWS (Sagemaker, EC2, S3, Lambda, API Gateway, DynamoDB, RDS, EKS); methods: projective transformation, neural networks.

Market state detection

Description: Based on historical prices of an asset (crypto or forex) determine the current general market state
Responsibilities:

• Design definition of market states;
• design algorithms for market state detection;
• deploy the system to work with live streams of data.

Technologies: SQL; Python; Numpy; Plotly, binance; Psycopg2; Scikit-learn; AWS (Sagemaker, EC2, S3, Lambda, API Gateway, DynamoDB, RDS, EKS); methods: linear regression.

Machine Learning Researcher

Jun 2019 - Jun 2022
Responsibilities:

• Researching and implementing computer vision techniques for eye-tracking, emotion recognition and attention measuring;
• implementing various statistical tests.

Technologies: Python; OpenCV; SQL; TensorFlow; PyMC3.

Machine Learning Engineer

June 2021 - June 2022
Responsibilities:

• Designing and implementing statistical testing for Reaction Time Testing;
• designing and implementing MaxDiff Analysis;
• improving a webcam based eye-tracking system;
• processing of eye-tracking results (smoothing, denoising, fixation and saccade detection).

Technologies: Python; OpenCV; SQL; TensorFlow; pymc3.

Machine Learning Engineer

June 2019 - June 2021
Responsibilities:

• Adapting eye-tracking system for mobile devices
• Developing an Emotion Detection system;
• developing an Attention Measuring system;
• data filtering of emotion and attention results (smoothing, outlier detection).

Technologies: Python; OpenCV; SQL; TensorFlow;

Research at Ulm University

March 2018 - November 2019
Responsibilities:

•  Implementing Heath–Jarrow–Morton framework for simulating/predicting interest rate curves;
•  implementing mortality rate prediction in an AM/EM setup.

Technologies: R; Monte Carlo simulations; Stochastic Processes; Statistics.

Lecturer at University

Sep 2020 - Present
Responsibilities:

• Teaching Machine Learning, Deep Learning, Computer Vision.

Education

Taras Shevchenko Kyiv National University, Ph.D., Machine Learning, Mathematics Ulm University, Master's degree in Actuarial Science

Certificates

• Deep Learning Specialization - Coursera
• Machine Learning - Coursera
• Finding structure in data - Coursera
• Training on labeled data - Coursera
• Math and Python for Data Analysis - Coursera
• Introduction to Deep Learning (with Honors) - Coursera
• TensorFlow in Practice Specialization - Coursera
• AI for Medicine Specialization - Coursera