Uygar Kurt

Uygar Kurt is an AI researcher and engineer who created this channel to share technical knowledge and personal projects. The content focuses on implementing and explaining machine learning models and techniques from scratch, covering topics such as GRPO, mHC, GLM-4.5, and Vision Language Models using PyTorch. It also includes deep dives into GPU programming with CUDA for operations like ReLU activation and vector addition through Tensara Solutions.

The channel provides tutorials on advanced AI engineering workflows, including benchmarking LLMs with multi-GPU support, fine-tuning models with HuggingFace, LoRA, and quantization, and implementing models like Llama 3 from scratch. Other technical topics covered are model merging with Mergekit, estimating the carbon footprint of ML models, and building conversational memory for LLMs using LangChain. The channel states it is a new and small platform for sharing in-depth technical AI research and engineering projects.