# Eren Can Almaz

> Eren Can Almaz is an Electrical Engineering (Elektrotechnik) student at RWTH Aachen University who builds interactive tools that make engineering visible — neural networks, flight control and signals. Open to HiWi / tutor roles in control, machine learning and signal processing. GitHub: https://github.com/writerforight

## Projects

- [Neural Nets Space Bending](https://writerforight.github.io/mlp-space-deformation/): interactive, browser-only visualization of how a small neural network (MLP) deforms space layer by layer and how backpropagation reshapes it. The MLP, backpropagation, SGD, Adam, Jacobians, PCA, the neural tangent kernel and continual-learning methods (replay buffer, orthogonal gradient descent, EWC) are written from first principles in plain JavaScript. Source: https://github.com/writerforight/mlp-space-deformation
- [Quadrotor Lab](https://writerforight.github.io/drone-wind-sim/): browser quadrotor simulator for controller design and tuning — 6-DoF rigid-body model integrated with RK4, motor model with lag and saturation, mixer, cascaded PID (position → attitude → body rate), manual flight and live gain tuning with step responses. Plain JavaScript. Source: https://github.com/writerforight/drone-wind-sim
- [Ders Transkript](https://github.com/writerforight/ders-transkript): offline lecture transcription desktop app for Windows, macOS and Linux (faster-whisper, PyQt6); turns lecture recordings and live microphone input into text without an internet connection.
- [specroots](https://writerforight.github.io/specroots/): browser dictionary that teaches English words through their Latin and Greek roots (spec, port, dict, duc, scrib …). Each word is split into prefix · root · suffix, glossed in Turkish, German, French, Italian and Spanish, and shown in example sentences with pronunciation; list and tree views, interface in six languages. Source: https://github.com/writerforight/specroots
- [YouTube Segment Downloader](https://github.com/writerforight/youtube-segment-downloader): PySide6 desktop app to download a full YouTube video or only a selected time range.
- [Char-Based-RNN](https://github.com/writerforight/Char-Based-RNN): character-level recurrent neural network that learns online, one step at a time, in pure NumPy with a Tkinter interface.

## Links

- [Personal page](https://writerforight.github.io/)
- [GitHub profile](https://github.com/writerforight)