Ibrahim Gülmez

Computer Engineer | AI/ML Specialist

Ankara, TR.

About

Highly motivated Computer Engineer specializing in Deep Learning, Computer Vision, and Full-Stack Development, backed by a strong academic record (GPA 3.85/4.0). Proven ability to design, implement, and deploy advanced AI models for classification, object detection, and segmentation, as demonstrated through impactful internships and a published graduation project. Seeking to leverage expertise in Python, PyTorch, and MERN stack to drive innovative solutions in AI/ML engineering or full-stack development roles.

Work

TUSAŞ A.Ş.

|

Internship

Ankara, , Türkiye

→

Summary

Led deep learning investigations for aircraft detection, applying advanced Convolutional Neural Networks and object detection models to enhance classification and interpretation.

Highlights

Conducted comprehensive exploratory data analysis on aircraft datasets, strategically selecting the Military AirCraft Detection dataset for optimized classification.

Implemented and optimized Convolutional Neural Networks (CNN) and EfficientNet B3 with transfer learning, achieving successful classification of aircraft detection data.

Applied Grad-CAM interpretation tools to enhance model explainability and interpretability for deep learning classification models.

Initiated and evaluated advanced object detection models, including R-CNN, Fast R-CNN, YOLO, YOLOv8, and DETR, for robust aircraft detection tasks.

AKGÜN YAZILIM

|

Internship

Ankara, , Türkiye

→

Summary

Conducted computer vision and AI research focused on brain tumor segmentation and classification, optimizing data pipelines and leveraging U-Net architecture.

Highlights

Executed comprehensive computer vision and AI research, focusing on brain tumor segmentation and classification.

Conducted thorough literature reviews and extensive exploratory data analysis to inform research on medical imaging.

Optimized data preprocessing techniques, enhancing the efficiency and quality of input for deep learning models.

Achieved promising results in brain tumor segmentation utilizing the U-Net architecture, accelerated by RTX 4090 GPU.

Enhanced academic writing skills through the preparation of research findings for publication.

Education

Ankara Yıldırım Beyazıt University

Ankara, , Türkiye
→
Bachelor's Degree Computer Engineering Grade: 3.85/4.0

Awards

First Place in Faculty of Engineering and Natural Sciences

Ankara Yıldırım Beyazıt University

Recognized for outstanding academic performance, achieving first place among peers in the Faculty of Engineering and Natural Sciences for the 2023-2024 academic year.

Publications

Automatic Recognition of Baby Crying Sounds

IEEE

Published research on deep learning techniques for automatic classification of infant crying sounds, presented at the 32nd Signal Processing and Communications Applications Conference (SIU).

Languages

Turkish

English

Certificates

AI Agents and Agentic AI

Coursera

Vector Databases for RAG

Coursera

Build RAG Applications

Coursera

Develop Generative AI Applications

Coursera

SIU 2024 (Presentation/Participation)

IEEE

Sky Discover 2023

TUSAŞ A.Ş.

Machine Learning

Analytics Vidhya

Applied Java Course

UDEMY

Skills

Programming Languages

Python, JavaScript, TypeScript, Java, C.

Machine Learning & Deep Learning

PyTorch, Hugging Face, Scikit-learn, TensorFlow (implied), Deep Learning, Computer Vision, Object Detection, Image Segmentation, Transfer Learning, Model Quantization, Domain Adaptation, Generative AI, RAG Applications, AI Agents.

Data Science & Analysis

Numpy, OpenCV, Pandas, SciPy, Matplotlib, Seaborn, Exploratory Data Analysis, Data Preprocessing.

Web Development

React, ReactNative, Node.js, Express.js, MongoDB, Flask, Full-Stack Development, API Integration.

Cloud & DevOps

Google Cloud Platform (GCP), Docker, Containerization.

High-Performance Computing

MPI, OpenMP, CUDA.

Tools & Methodologies

Grad-CAM, U-Net Architecture, YOLO, DETR, R-CNN, Fast R-CNN, EfficientNet.

Projects

Automatic Recognition of Baby Crying Sounds (Graduation Project)

→

Summary

Developed and deployed a deep learning-based system for real-time automatic classification of infant crying sounds, designed to identify basic needs. The project involved evaluating ANN and CNN models, deploying the most successful model on Google Cloud Platform (GCP) with Flask and Docker, and developing a React Native mobile application with a Node.js, Express, and MongoDB backend. Research findings were published and presented at SIU 2024.

Deep Learning: Domain Adaptation

→

Summary

Developed and evaluated adversarial domain adaptation models using gradient inversion and ADDA approaches on the Modern-Office-31 dataset. Achieved significant improvement in target domain accuracy from 85% (source only) to over 90% with adversarial training.

Deep Learning: Quantization

→

Summary

Demonstrated model compression using specialized quantization techniques (int8 and int4), exploring methods like asymmetric, symmetric, channel-wise, and group-wise quantization. Featured a custom linear layer quantizer and 4-bit approach to reduce model size, achieving significant reductions in memory footprint and inference time while largely preserving original performance.

Full-Stack Web Project: VOCABA

→

Summary

Developed VOCABA, a dictionary-like application utilizing a free API for word search. Users can save learned words, add them to learning processes, and upload word-related videos for pronunciation. Built using the MERN tech stack (MongoDB, Express, React, Node.js).