Emircan Çağlar (Emircan Caglar) is a machine learning engineer working in deep learning, reinforcement learning, and generative models. This site is his portfolio: a scroll-descent through a colossal machine cathedral, modeled in Blender, rendered in the browser with three.js, and scored with generative Web Audio.
Emircan Çağlar, derin öğrenme, pekiştirmeli öğrenme ve üretken modeller üzerine çalışan bir makine öğrenmesi mühendisidir. Bu site onun portfolyosudur: Blender'da modellenmiş, tarayıcıda three.js ile çalışan, makine katedralinin içinden kalbine inen etkileşimli bir WebGL deneyimi.
How to explore: scroll to descend through the cathedral. Near the core, his projects stand in a procession of plinths — hover a plinth to see the project’s name, and click it to open the project’s record. The full project list also appears below.
Diffusion model that preserves topological invariants during the denoising process. Generates manifolds with guaranteed genus and Betti numbers.
Python, PyTorch, Topology, Diffusion Models
Source on GitHubCustom RL environments and agents for continuous control tasks. PPO and SAC implementations with real-time visualization.
Python, PyTorch, Gymnasium, RL
Source on GitHubLow-latency audio analysis pipeline for live spectral decomposition, beat detection, and feature extraction at 60fps.
TypeScript, Web Audio API, FFT, WebGL
Source on GitHubA scroll-descent through a colossal machine cathedral, down to its burning core. Every asset modeled in Blender and streamed as glTF; the scroll itself cranks the machinery awake. Generative sub-bass score in Web Audio.
Blender, Three.js, TypeScript, glTF, WebAudio
Source on GitHub