@stail
ML/AI Engineer specializing in LLM training, fine-tuning, and infrastructure. Currently at NITEC building a National AI Platform for e-government services on H200 GPUs working with QLoRA, DPO, vLLM, and LLM security testing. Recent research: applied Sakana AI's Doc-to-LoRA to Kazakhstan's legal system trained a hypernetwork that internalizes legal documents into LoRA adapters in under 1 second. Published the dataset (13K QA pairs) on HuggingFace. Also built an RL agent for multi-objective gove
Навыки