XAI Dashboard for Deep Learning
Interactive web dashboard for explaining deep learning model predictions using SHAP, LIME, and Grad-CAM visualizations.
Computer Vision & Deep Learning Engineer | Image Processing, LLMs, Passionate about ADAS & Autonomous Systems 🧠 🤖 🏎️ 🏁
I am an R&D Software Engineer with an MSc in Computer Engineering, driven by an innate curiosity to understand "how things work" under the hood. My career is defined by a hybrid focus: conducting rigorous academic research and translating those findings into functional software prototypes.
Current Focus: AI & R&D Management At ReklamStore, I play a key role in the R&D department. My responsibilities extend beyond coding; I manage the full innovation lifecycle—from conducting literature reviews and drafting patent applications to developing Python-based APIs that power AI and Computer Vision features. I actively contribute to architectural decisions and ensure our projects meet high academic standards for TÜBİTAK reports.
Core Passion: Autonomous Systems While my current role focuses on Generative AI, my engineering foundation is built on ADAS and Autonomous Vehicles. During my graduate studies and time with the Alatay Autonomous Team, I gained hands-on experience with C++, ROS, and Simulation Environments (AirSim, Unreal Engine 4, TORCS). I specifically worked on Deep Reinforcement Learning agents and Explainable AI (XAI) to make self-driving decisions transparent.
Goal I am looking for a role where I can leverage my R&D discipline, backend skills, and engineering background to contribute to the next generation of automotive technologies.
ReklamStore İnternet Pazarlama A.Ş. - İstanbul
I play a key role in "CreabitAI," a project supported by the TÜBİTAK-1501 program. I bridge the gap between academic research and commercial product development. I design NLP microservices using OpenAI API and build video processing pipelines with FFMPEG and Python, and MoviePy. I am responsible for the official reporting process. I prepare detailed technical reports and present our progress to TÜBİTAK referees and academic boards to ensure successful project continuation.
Pamukkale University
Explainable Artificial Intelligence (XAI) for Deep Reinforcement Learning-Based Autonomous Driving. Focused on solving the "black box" problem in autonomous vehicle decision-making. I developed racing agents using DRL algorithms (TD3, DDPG) in the TORCS simulator and implemented SHAP analysis to make their decisions transparent and interpretability. My thesis demonstrated that TD3 outperformed PPO/DDPG in TORCS, and SHAP validated that the agent's behavior aligned with human-like driving patterns. This work contributes to safer and more trustworthy autonomous systems by providing insights into the decision-making process of DRL agents.
Oncostone LTD. - Denizli
I led the R&D department of a Technopark startup, managing the full lifecycle of an AI-driven Enuretic Alarm Device designed to treat pediatric bedwetting. I was responsible for the entire project management process, including the preparation of official R&D reports for the Technopark administration. Technically, I developed algorithms to analyze the child's sleep data from the moment of falling asleep up to the REM stage, aiming to predict high-risk intervals. I developed the firmware for ESP32 using C++ (ANT+ protocol) for low-power communication and wrote Python algorithms to process sleep signals and detect wetness events.
Alatay Mobil Autonomous Team
I was a core R&D member of the university racing team (Ranked 4th in Teknofest 2020). I was responsible for the software stack, from simulation to the real road. I developed lane detection and path planning in ROS, integrated LiDAR and cameras on Nvidia Jetson, and managed CAN Bus communication.
Pamukkale University
Focused on Image Processing, Computer Vision, Deep Learning, Embedded Systems, and Autonomous Vehicles. Supplemented core curriculum with intensive research in autonomous vehicle simulation and control logic.
Selected works from my portfolio
Interactive web dashboard for explaining deep learning model predictions using SHAP, LIME, and Grad-CAM visualizations.
Kubernetes-native machine learning training and deployment pipeline with auto-scaling, experiment tracking, and model versioning.
Commands and snippets I use daily
# System update
$ sudo apt update && sudo apt upgrade -y
# Install Python env
$ python3 -m venv .venv
$ source .venv/bin/activate
# Check system resources
$ htop
# Launch navigation stack
$ ros2 launch nav2_bringup navigation_launch.py
# Build workspace
$ colcon build --symlink-install
# Source workspace
$ source install/setup.bash
# List available nodes
$ ros2 node list
# Build and run containers
$ docker-compose up -d --build
# View running containers
$ docker ps --format "table {{.Names}}\t{{.Status}}"
# View logs
$ docker-compose logs -f
# Stop all containers
$ docker-compose down
# Initialize repository
$ git init
# Create feature branch
$ git checkout -b feature/new-feature
# Stage and commit changes
$ git add . && git commit -m "feat: add new feature"
# Push to remote
$ git push origin feature/new-feature
# Install requirements
$ pip install -r requirements.txt
# Run training script
$ python train.py --epochs 100 --batch-size 32
# Start Jupyter notebook
$ jupyter notebook --no-browser --port=8888
# Export trained model
$ python export_model.py --format onnx
# Get all pods
$ kubectl get pods -A
# Apply configuration
$ kubectl apply -f deployment.yaml
# View pod logs
$ kubectl logs -f pod-name
# Scale deployment
$ kubectl scale deployment app --replicas=3
Technical articles and notes
Welcome to my blog! In this first post, I'll walk through how I built this Markdown-based blog system using Flask and Python.
Learn how to structure your Flask templates with Jinja2 inheritance for clean, maintainable code.
Guide how to pair your bluetooth devices both OS in same device.
What I do when I'm not coding...
I follow every race weekend not just as a fan, but with an engineer's eye. I enjoy analyzing telemetry data and tracking aerodynamic upgrades. I deeply admire Adrian Newey's engineering genius. Occasionally, I use Python to visualize race pace and strategy data myself.
Let's Collaborate
Istanbul, Türkiye 🇹🇷
Open to Full-time R&D Roles & Technical Consulting