IT engineer
UiT The Arctic University of Norway
Taras Andrushko is an accomplished AI Engineer with expertise in Norwegian Large Language Models (LLMs) and automatic speech recognition (ASR). Holding a Master’s degree in Language Technology from the University of Oslo, he played a key role in developing the preliminary interface for the Norwegian Constructicon.
With extensive hands-on experience in speech processing, text-to-speech (TTS), and ASR research, Taras excels in designing and implementing end-to-end machine learning pipelines — from data preprocessing and model training to evaluation and deployment. His technical proficiency spans cutting-edge tools and frameworks, including PyTorch, Hugging Face, LoRA fine-tuning, transformers, GANs, diffusion models, and speech tokenizers.
In addition, he is adept at deploying models in production environments, leveraging technologies such as Docker, Kubernetes, and ONNX/TorchScript. Driven by a passion for innovation, Taras is dedicated to creating natural, expressive speech synthesis and advancing voice AI to power global applications.
