Zeki Bilgin - Senior Data Scientist

İstanbul, Turkey

Experience

Jan 2025 - Dec 2025

1 year

United Kingdom

Conducted research and development on weight biomarkers from facial images for studies related to eating disorder treatments.
Implemented feature extraction algorithms using MediaPipe and facial landmarks to build machine learning models for weight and BMI (Body Mass Index) prediction from face images.
Developed CNN-based and transformer-based computer vision models for predicting BMI variations from image pairs.
Applied model explanation techniques such as saliency maps and Grad-CAM to improve interpretability of deep learning models.

Jan 2024 - Dec 2025

2 years

United States

Researched, developed, and implemented a palm-based biometric identification system.
Implemented and fine-tuned Siamese neural networks for palm-based biometric identification (image similarity), and trained and tested them.
Developed anti-spoofing algorithms based on neural networks and classical image processing techniques.
Created image segmentation models for background replacement.

Jan 2020 - Dec 2024

5 years

İstanbul, Turkey

Researched, developed and implemented AI/ML and deep learning algorithms with a focus on computer vision applications for white-good and home appliances, including object recognition (YOLO), image classification (CNN), cooking termination (CNN-LSTM), human detection, image undistortion, image blurry measurement, and other intelligent applications for image enhancement, correction, cropping and analysis.
Implemented NLP models, including one for ranking customer comments on a scale from 1 to 5 and another for text segmentation.

Jan 2018 - Dec 2020

3 years

İstanbul, Turkey

Conducted research in Ericsson’s business areas including software vulnerability prediction from source code by machine learning, 5G and IoT with a focus on security and privacy aspects, and IoT use cases such as Smart Grid and Smart Manufacturing (Industry 4.0) including functional safety, integrating AI into these systems.
Implemented privacy-preserving techniques for machine learning such as federated learning, differentially private LSTM for load forecasting, and secure aggregation by multiparty computation.
Filed multiple patent applications related to the mentioned technologies and 3GPP standards.

Jan 2017 - Dec 2018

2 years

United States

Developed mathematical models and performed simulation experiments for the ODIN (Open Dynamic Interaction Networks) project to analyze information diffusion and social influence in social networks by implementing machine learning techniques.

Ai/ml
Data Science
Deep Learning
Machine Learning
Computer Vision
Image Processing
Nlp
Llms
Cybersecurity
Privacy Preserving Technologies
Wireless Communication
Very Proficient With Deep Learning And Machine Learning Algorithms (E.g. Supervised & Unsupervised Algorithms, Reinforcement Learning, And Different Neural Network Architectures Such As Cnn, Lstm, And Siamese, As Well As Typical Ml Algorithms Such As Svm, Decision Trees, Knn, Rf, And More – Have Industry Experience And Publications On These Areas)
Very Proficient With Computer Vision Techniques Such As Object Recognition (By Yolo), Image Classification (By Cnn), Image Segmentation, Image Similarity Etc.
Very Proficient With Privacy-preserving Ml Solutions Such As Federated Learning, Differential Privacy, Multiparty Computation (Have Publications On This Issues)
Very Proficient With Nlp, And Customization And Fine-tuning Of Llms
Very Proficient With Programming In Python, C And C++
Very Proficient With Tensorflow, Pytorch And Scikit-learn Machine Learning Frameworks
Very Proficient With Image Processing Techniques By Opencv
Very Proficient Aws Services
Experience With Postgresql, Fastapi, And Containerization
Very Proficient With Privacy Preserving Technologies On Data Processing
Very Proficient With Ns3 Network Simulator