I develop software-driven engineering systems spanning backend applications, artificial intelligence, computer vision, embedded computing, autonomous robotics and industrial automation. My work connects software architecture and applied AI with hardware-aware deployment and practical system integration.
My engineering path began with programming, numerical methods, data analysis and applied machine learning. It has expanded through full-stack systems and edge AI into ROS 2 mobile robots, embedded Linux and production-oriented industrial software.
I focus on clear interfaces, testable behavior, operational constraints and reliable integration between software services, AI models, devices and physical systems. Earlier software, data and AI projects remain part of this profile because they provide the foundation for my current robotics and automation work.
| Software Engineering | Artificial Intelligence & Computer Vision |
|---|---|
| C++, Python, Rust and TypeScript · Backend services and REST APIs · Software architecture · Automated testing · Linux, Docker and CI/CD · Full-stack applications Project evidence: MacroTR, industrial orchestration services, local-first desktop/service systems and C++ development tooling |
Machine learning · OpenCV, YOLO and ONNX · PyTorch and TensorFlow Lite · Audio AI · Edge inference · Evaluation and deployment workflows Project evidence: Hailo benchmarking, C++ zone analytics, audio classification and edge-vision systems |
| Robotics & Industrial Automation | Embedded Systems & Edge Computing |
| ROS 2, Gazebo and RViz · SLAM Toolbox, AMCL and Nav2 · Autonomous mobile robots · Mission and workflow orchestration · Robot fleet integration · Siemens S7 and OPC UA integration · PROFINET and Modbus TCP diagnostics | Raspberry Pi and Hailo accelerators · ESP32 · Embedded Linux · Device provisioning and networking · Sensor and video pipelines · IPC and shared memory · Real-time and streaming data processing |
The descriptions below are scoped to audited repository evidence. For collaborative or commercial systems, they describe the engineering work represented in my portfolio without claiming sole authorship.
| Project | Engineering problem, contribution and approach | Status |
|---|---|---|
| Industrial Orchestration Platform | Coordinates recoverable workflows across industrial controllers, robot systems and operator interfaces. The represented work includes a C++20 service architecture, Siemens S7/OPC UA/HTTP adapters, PostgreSQL, RabbitMQ, a React operator UI, automated tests, recovery behavior and cross-platform CI. | Private commercial work · Sanitized summary |
| ROS 2 Mini AMR | Implements an autonomous mission lifecycle from robot description and simulation through mapping, localization and navigation. Uses URDF/Xacro, Gazebo, RViz, SLAM Toolbox, AMCL/Nav2, custom ROS 2 interfaces, mission state management, SQLite history and automated tests. | Private personal project · Active development |
| MacroTR | Turns public Turkish macroeconomic data into a maintainable full-stack platform. Uses FastAPI, React/TypeScript, PostgreSQL, Redis, Docker, backend/frontend tests and GitHub Actions around TCMB EVDS data. | Public · Active development · CI verified |
| Hailo Edge-AI Benchmarking | Examines CPU-versus-NPU inference behavior on Raspberry Pi 5 and Hailo hardware. Includes benchmark scripts, committed CSV measurements, charts and analysis for throughput, latency, utilization and temperature. | Public · Benchmark study · Reproducibility improvements planned |
| Zone Analysis | Builds a real-time video-analysis pipeline beyond a notebook prototype. Combines C++17, OpenCV, YOLO-ONNX, RTSP/video input, polygon zones, temporal tracking, dwell events, snapshots and CSV output. | Public · In development · Tests and demo still needed |
| Audio AI Detector | Investigates cry-versus-non-cry audio detection and reason classification without hiding weak results. Uses leakage-aware splits, a two-stage pipeline, evaluation reports and tests; the documented reason-classification limitation remains an open data problem. | Private research prototype · Experimental, not a medical system |
| Local-First AI Workspace | Integrates desktop workflows, retrieval and document/spreadsheet processing while keeping the primary user experience local. The audited architecture combines C++20/Qt, Python services, orchestration and a substantial automated-test suite. | Private product work · Sanitized summary |
| C++ Environment Bootstrap | Reduces setup drift in cross-platform C++ development through repeatable environment scripts, documented toolchain checks and CI workflows. | Public · Developer tooling |
Private work remains private. Customer identities, production configurations, network details, credentials, proprietary workflows, restricted data and source code are intentionally excluded.
- Software systems: C++, Python, Rust and TypeScript across backend services, desktop applications, operator interfaces and engineering tools, with relational data stores, messaging and automated delivery workflows.
- Artificial intelligence: computer-vision and audio pipelines covering dataset preparation, evaluation, failure analysis, ONNX/TensorFlow Lite deployment and hardware-accelerated inference.
- Autonomous robotics: personal ROS 2 development using simulation, robot descriptions, mapping, localization, Nav2 actions, custom interfaces and mission-level state management.
- Industrial integration: commercial engineering work involving adapter-based communication, PLC-facing systems, mobile-robot and fleet interfaces, workflow recovery, operational diagnostics and operator tooling.
- Embedded and edge systems: Raspberry Pi, Hailo, ESP32 and embedded Linux work involving device networking, provisioning, video streams, local APIs and hardware/software integration.
- Engineering analysis: tested tools for AMR fleet sizing, throughput, CAPEX/OPEX, ROI and reference-model comparison, with assumptions and limitations kept explicit.
| Engineering purpose | Technologies used in audited projects |
|---|---|
| Languages | C++, Python, Rust, TypeScript, SQL, Dart, MATLAB |
| Robotics & simulation | ROS 2, Nav2, SLAM Toolbox, AMCL, Gazebo, RViz, URDF/Xacro |
| Industrial systems | Siemens S7, OPC UA, industrial HTTP APIs, PROFINET and Modbus TCP service diagnostics |
| AI & computer vision | PyTorch, TensorFlow, scikit-learn, OpenCV, YOLO, ONNX, TensorFlow Lite, HailoRT |
| Backend & data | FastAPI, PostgreSQL, SQLite, Redis, RabbitMQ, REST, WebSocket |
| Systems & delivery | Linux, Docker, CMake, GitHub Actions, GStreamer, shared memory and IPC |
| Interfaces & devices | React, Qt, Flutter, Raspberry Pi, ESP32, Hailo-8, RTSP cameras |
- Programming foundations — numerical methods, signal processing, data analysis and small desktop or command-line applications.
- Software and applied AI — backend APIs, full-stack systems, machine learning, audio classification and computer-vision pipelines.
- Edge deployment — Raspberry Pi, embedded Linux, Hailo acceleration, device networking and hardware-aware inference benchmarking.
- Robotics and industrial systems — ROS 2 navigation, autonomous mobile robots, fleet engineering, industrial communications and resilient orchestration software.
This progression is presented as an engineering continuum: software provides the architecture, AI provides perception and decision support, embedded systems connect computation to devices, and robotics and automation bring those capabilities into physical operations.
My academic background combines Electrical and Electronics Engineering with Economics. Engineering is the primary foundation of my work; economics informs selected projects involving data platforms, quantitative models and investment analysis.
Current technical interests include autonomous navigation, edge-AI benchmarking, reliable industrial orchestration, embedded device lifecycle design and validation of engineering calculation models.
- Technical writing: Medium
- Engineering demos and project videos: YouTube
- Professional contact: LinkedIn
- Source code and project history: GitHub
Project descriptions reflect repository evidence reviewed in November 2026. Private and commercial systems are represented only through sanitized, non-confidential summaries.