Resume

View PDF Version

Personal Info

Name
Berkin Deniz Kahya
Title
AI Researcher, Software Engineer
Email
berkindeniz2000@gmail.com
Location
İstanbul, Turkey
Links
LinkedIn GitHub
Summary
AI Researcher specializing in medical image processing and segmentation. Currently working on application ranking/sorting tasks with Huawei.

Work Experience

AI Researcher
02/25 - Present
Huawei
  • --
AI Researcher & Engineer
10/23 - 10/24
Mogram
  • Mass & Tumour & Calcification segmentation with various encoder-decoder networks.
  • Used patch-based transformer encoder-decoder network to enhance f1 and box f1 scores by around 10%.
  • R&D of the segmentation training pipeline with MLflow.
  • Combined fast-slic algorithm with adaptive thresholding to gain roughly 10x speed over CPU-bound processes.
  • Design of on-premise service authentication and encryption routines.
  • Anonymization of the DICOM file structure and organization of datasets.
  • Resolved Docker and MongoDB issues in a remote setup.
Team Member at Inzva
02/21 - Present
Inzva
  • Design and tracking of learning materials as a member of the AI Team.
  • Gave lectures on Structures of ML projects and Computer Vision Fundamentals.
  • Google Tensorflow Developer & DL Specialization certificates.
  • Worked with Mamba architecture on State Space Tracking/Modelling of Transformer-based LLMs.
  • Application of various deep learning models and study programs to solve real-life and designed challenges for peer learning.
AI Research Student
06/22 - 10/23
ASIL Laboratories
  • Dataset creation, labelling & preprocessing of medical histopathology images.
  • Segmentation for pixel-wise tumour detection, f1 over 70% with unified resampling for post-processing.
  • Image augmentation techniques with Autoencoder and Diffusion models.
  • Publishing a paper to JID Q1 is in progress.
Machine Learning Intern
07/22 - 08/22
Optiwisdom
  • Integration of the customer database into NoSql.
  • Various ML methods (boosting & tree algorithms) for revenue prediction.

Education

Mechatronics Engineering, Yıldız Technical University
09/19 - 12/24
GPA: 3.04/4.00

Projects

Other Study

  • Worked on 3D car localization with sensor fusion and YOLO
  • Building an autonomous microscope to classify fungi cells for the graduation project.
  • Deep Learning Specialization courses and certificate by Andrew Ng.
  • Tensorflow Developer courses and certificate by Laurence Moroney.

Self Study Projects

  • Basic chess game and a user interface using Python.
  • Worked on creating a chess bot using alpha-beta pruning.
  • Visualizing signals using Laplace Transformation with manim library on Python.
  • Training of UNet/SwinUnetR architecture on the Skin Lesion Segmentation dataset.

Skills

Tools and Languages
Python Git PyTorch TF & Keras OpenCV Cuda MLflow Docker MongoDB C & C++ Matlab SQL
Research Interest & Areas
Medical Image Processing & Segmentation Scene generation in motion pictures Mathematical Modeling and Visualization

Languages

Communication
Turkish English

Interests

Other Interests

  • Mathematical modelling and curious about its applications in learning.
  • Reading articles about new topics related to up-to-date methods in AI.