Available for research & collaborations

Necip
Türkmenoğlu

Building intelligent systems at the intersection of computer vision and deep learning. Currently studying at Konya Food and Agriculture University — crafting real-world AI solutions and exploring the boundaries of lightweight neural architectures.

View My Work Get in Touch
99.05%
Classification Accuracy
1 Paper
Published Research
41,793
Images Trained On
7.06ms
Inference Time
CURRENT STATUS
🎓 BSc Computer Eng.

about_me.md

Turning Data Into
Intelligent Systems

I'm Necip Türkmenoğlu, a Computer Engineering student at Konya Food and Agriculture University with a deep passion for building AI systems that work in the real world — not just on benchmarks.

My research focuses on lightweight convolutional neural networks for embedded deployment. I co-authored a peer-reviewed paper benchmarking MobileNetV3 Small, ResNet34, and SqueezeNet1.1 for real-time driver drowsiness detection — achieving 99.05% accuracy at just 7ms inference time on a 5.93MB model.

Beyond academia, I build software tools, experiment with web technologies, and maintain this studio — Studio505 — as my personal space to ship projects and explore ideas at the edge of AI and software engineering.

Computer Vision Deep Learning PyTorch Embedded AI Transfer Learning Research Python Web Dev

projects.json

Research & Projects

Real-world systems built with purpose — from published research to side projects.

⭐ Featured Research Paper 📄 Published

Real-Time Driver Drowsiness Detection via Lightweight CNN Benchmarking

A comprehensive performance assessment of pre-trained CNN architectures (MobileNetV3 Small, ResNet34, SqueezeNet1.1) for real-time driver drowsiness classification using the Driver Drowsiness Dataset (41,793 images). MobileNetV3 Small emerged as the optimal architecture — delivering state-of-the-art accuracy with a compact 5.93MB footprint suitable for edge devices.

99.05% Accuracy
99.13% F1-Score
99.31% Recall
~7ms Inference Time
5.93MB Model Size
41,793 Training Images
Python PyTorch MobileNetV3 ResNet34 SqueezeNet1.1 Transfer Learning Google Colab T4 GPU Computer Vision BCEWithLogitsLoss
Request Paper GitHub

Studio505 — Personal Domain

Configured a full production domain stack on Cloudflare — custom DNS, Cloudflare Pages deployment with GitHub CI/CD, www redirect rules, and Cloudflare Tunnel for local service exposure.

LIVE Visit ↗
Cloudflare DNS Pages GitHub Actions

MCP / Codex Local Bridge

Running a local AI development infrastructure using Cloudflare Tunnel to securely expose local MCP and Codex services through mcp.studio505.pp.ua without exposing home IP.

IN PROGRESS
Cloudflare Tunnel Windows 11 MCP

Future Research

Upcoming work: integrating temporal modeling (LSTM, Vision Transformers) for video-based drowsiness detection. Extending toward multimodal fatigue indicators including EEG and head pose.

PLANNED
Vision Transformer LSTM EEG Embedded AI

skills.yaml

Technical Stack

The tools and technologies I work with to build real-world AI and software systems.

AI / Machine Learning

PyTorch 92%
Transfer Learning / Fine-tuning 88%
CNN Architecture Design 85%
TensorFlow / Keras 75%

Computer Vision

Image Classification 95%
OpenCV 82%
Facial Landmark Detection 80%
Object Detection (YOLO) 72%

Programming & Tools

Python 93%
Git & GitHub 86%
JavaScript / HTML / CSS 75%
Data Analysis (NumPy, Pandas) 82%

Infrastructure & DevOps

Cloudflare (DNS, Pages, Tunnel) 85%
Google Colab / GPU Training 88%
Linux / Windows Dev Environment 78%
Embedded Systems (Jetson Nano) 68%

contact.ts

Let's Connect

Whether you're interested in research collaboration, have a project idea, or just want to talk AI — my inbox is open. I'm currently open to new opportunities and research partnerships.

Currently Available

Open to research collaborations, internship opportunities, and interesting AI/ML projects. Based in Turkey 🇹🇷 — remote-friendly.

Research Collaboration Internships AI Projects Open Source Remote

// quick note

My research paper on driver drowsiness detection is available on request. Feel free to reach out if you're working in safety-critical AI or embedded vision systems.