The theme ships in two primary variants: Monolith Classic (pure neutral grays) and Monolith Black (true #000000 pitch black for OLED displays). Each variant supports both light and dark modes, matching SVG folder icons, and consistent code editor syntax palettes.

The repository includes a standalone Bash installer and a lightweight monolith-theme command-line utility for toggling variants and managing updates directly from the terminal. For automated installer scripts and usage flags, see details in the GitHub repository.

This extension rounds all application windows automatically to match the surrounding desktop. Instead of requiring manual configuration sliders, it inspects the live GTK theme stylesheet to extract the border radius, hairline outline, and drop shadows. The clipping logic hooks into Mutter window actors without keeping continuous render loops active, so idle windows consume no extra GPU or CPU resources.

For installation instructions, configuration options, and setup guides, see details in the GitHub repository.

The architecture separates concerns into three layers: pure data (YAML declarations, Markdown entries, and static media), an engine compiler that validates inputs against strict JSON Schemas (additionalProperties: false), and a theme layer that renders accessible, responsive HTML and minified CSS.

Key features include automated bidirectional taxonomy indexing for declared skills, full-text client-side search index compilation, RSS/Atom generation, and automated GitHub Pages deployment triggered via repository dispatches. The codebase is thoroughly tested with automated Vitest unit and integration suites. For architecture documentation, schema specifications, and build workflows, see details in the GitHub repository.

In the Spring 2025-2026 semester, I earned a 4.00 GPA across 30 ECTS. This entire final term was dedicated to Vocational Training in Business (İşletmede Mesleki Eğitim), spent as a full-time software engineering intern on the mobile team at Orion Innovation. The grade reflects fifteen weeks of shipping production Kotlin code, integrating real-time WebRTC calling, building offline-first database flows, and resolving architecture debt in an enterprise messaging application.

The coursework that paid the highest dividends was systems programming, operating systems, and computer networks. Understanding low-level hardware constraints, memory allocation, and concurrency made debugging real-world production systems during my engineering internship at Orion Innovation much more intuitive.

During my senior year, that foundation supported both shipping real-time communications software and conducting capstone research on deep-learning models for medical imaging. The five semester distinctions earned throughout my studies reflect that steady focus across each academic term.

In Fall 2025-2026, I earned a 3.88 GPA across 30 ECTS, my highest classroom semester. The coursework directly aligned with my practical goals: Mobile Programming, Introduction to Artificial Intelligence, and Applied Unsupervised Learning Methods. This combination provided the immediate technical runway for the Android engineering internship at Orion Innovation that began the following week, as well as advancing my senior research in computer vision.

Rather than relying on userscripts or JavaScript-driven styling extensions that can cause flashes of unstyled content, Pitch Black is built as a lightweight, native browser extension. It injects a pure CSS stylesheet at document_start across Chrome, Firefox, and Safari.

With no background scripts, no analytics, and zero runtime JavaScript execution, the theme renders instantaneously without impacting page load performance. Available on the Chrome Web Store and Firefox Add-ons, or see build instructions in the GitHub repository.

The benchmark compares ten distinct convolutional neural network configurations, including transfer learning backbones (DenseNet169, DenseNet121, InceptionResNetV2, ResNet50), edge-oriented models (MobileNetV2, NASNetMobile), and a custom convolutional network (OzNET) trained from scratch. DenseNet169 produced the leading result, reaching 93.91% test accuracy, a 0.9388 weighted F1-score, and a Cohen’s Kappa of 0.9344.

The full pipeline covers data normalization, augmentation, learning rate scheduling, confusion matrices, and ROC-AUC curves, packaged into a single reproducible notebook designed for execution on GPU-accelerated environments. For experiment notebooks and comprehensive metric breakdowns, read more in the GitHub repository.

Built with modern Android practices, the application uses Jetpack Compose for declarative UI, MVVM architecture, and Kotlin Coroutines/Flow for reactive data management. Firebase Authentication and Firestore provide real-time updates and role-based access control, distinguishing between regular campus reporters and administrative triage teams.

Key features include an interactive Google Maps interface displaying categorized incidents, photo attachments for damage reports, and an administrative dashboard for status tracking and urgent campus broadcast notifications. For architectural documentation, app permissions, and build instructions, see details in the GitHub repository.

Understanding the terminal and command-line Git makes working in team repositories much more reliable. Clean branch hygiene, meaningful commit messages, and structured pull requests make collaboration straightforward.

Part of Meta’s Android Developer specialization, alongside Introduction to Android Mobile Application Development.

The curriculum spans seven core areas: adversarial search (minimax with alpha-beta pruning), propositional logic, probability models (Bayesian networks, Markov models), constraint satisfaction, reinforcement learning (Q-learning), neural networks with TensorFlow, and natural language processing.

Implementing these algorithms by hand provided a solid mental foundation for diagnostic machine learning. It makes it much easier to reason about optimization dynamics, loss functions, and data representation in production models.

All twelve projects and documentation are available in the GitHub repository.

Working through these platform basics gave me a reliable mental model of the Android ecosystem before starting my internship at Orion Innovation. Understanding Gradle dependencies, layouts, and resource structuring made navigating a large enterprise codebase much more intuitive.

The hands-on project involves setting up and managing a public repository with real collaborative constraints. It reinforced treating pull requests and commit messages as primary tools of communication, which was directly useful when coordinating cross-platform features with other developers.

The simulation generates sphere meshes textured with NASA imagery for each major planet and satellite. Orbital dynamics are handled through hierarchical model-view-projection matrix transformations, accounting for axial tilt, rotation speeds, and moon-planet-sun parent-child relationships. The lighting shader implements Phong illumination with the Sun functioning as an emissive light source, set against a cubemap starfield skybox.

Navigation is driven by a 6-DOF free-fly camera with variable speed controls and FOV-scaled zoom. A Dear ImGui overlay provides real-time simulation metrics and timing adjustments. For compilation instructions and camera controls, see details in the GitHub repository.

The application provides dual user interfaces: an interactive command-line interface (CLI) for headless or fast terminal operation and a Java Swing desktop GUI. It manages room availability across multiple tiers (Single, Double, Suite), processes customer bookings, tracks check-ins, and enforces reservation conflict detection to prevent overbooking.

Data is persisted through SQLite using standard JDBC prepared statements to guard against injection. The project is managed via Maven with automated unit test suites verifying booking business rules. For build instructions and usage guides, read more in the GitHub repository.

Core topics include precision text objects (manipulating blocks inside parentheses, quotes, or tags), regex search and substitution, macros for repetitive edits, split windows, and buffer management. It also covers configuring Vim and integrating terminal tools into a direct terminal workflow.

Investing the time to master modal editing makes writing and refactoring code much faster, especially when operating over remote SSH sessions or inside headless servers.

The gameplay centers on defending against escalating enemy waves. The game loop maintains a fixed 60 FPS update cycle, managing player state, projectile trajectory physics, collision bounding boxes, and dynamic particle explosions. Enemy units feature different attack patterns and weapon spreads, concluding in multi-phase boss encounters.

The project includes custom sound synthesis, animated UI states (menu, active play, pause, game over), and local persistence for high-score leaderboards. For gameplay controls, asset credits, and setup instructions, see details in the GitHub repository.

The curriculum covers fundamental algorithms and data structures, including sorting routines, linked lists, hash tables, tries, and binary search trees. Later weeks transition into Python, relational database design with SQL, and building interactive web services. Each problem set is tested for computational correctness, edge cases, and memory safety with Valgrind.

Understanding the underlying memory model removes the mystery from higher-level runtime environments and makes debugging complex performance problems far easier.

All problem sets and my final project are documented in the GitHub repository.

The collection bridges low-level systems programming and high-level artificial intelligence:

• CS50x (Introduction to Computer Science): Focuses on low-level memory layout, pointer manipulation, and foundational data structures (linked lists, hash tables, tries) in C, alongside web development in Python and SQL.
• CS50 AI (Artificial Intelligence with Python): Explores algorithmic decision-making, covering graph search algorithms, logic and inference engines, probability models, reinforcement learning, and neural network classification with TensorFlow.

All problem sets include test suites, design notes, and source code organized into modular directories. For complete solution directories and coursework notes, read more in the GitHub repository.