Theta Labs, Kangwon National University join hands to advance AI with EdgeCloud

Theta Labs expands AI research partnerships as Kangwon National University joins Theta EdgeCloud’s decentralized GPU network.

SAN FRANCISCO, CA – March 5, 2025 – Theta Labs, a global leader in decentralized cloud infrastructure for AI, media, and entertainment, announces its latest academic customer, Kangwon National University Data Analytics & Machine Intelligence (DAMI) Lab, led by Professor Kyeongpil Kang. 

Bringing on board Kangwon National University as Theta EdgeCloud’s 23rd AI and video customer marks a significant milestone, collectively achieving over 85% GPU utilization rate across its worldwide network and over 230 million TFUEL locked by Elite Booster node operators. 

Theta EdgeCloud is proud to power a number of leading enterprises such as NHL’s Las Vegas Knights and global esports teams, including FlyQuest and Evil Geniuses, and top-tier academic institutions, Seoul National University, KAIST, University of Oregon, Michigan State University, and Singapore’s renowned research institution NTU, expanding its role in advancing AI research and real-world applications. 

About Professor Kyeongpil Kang and DAMI Lab

Professor Kyeongpil Kang is an Assistant Professor in the Department of Computer Science and Engineering at Kangwon National University, specializing in natural language processing (NLP), machine learning, and data mining. He earned his Ph.D. in Computer Science from Korea University, where he studied under Professor Jaegul Choo, and later worked as a machine learning researcher at SK Telecom AI Center, Scatter Lab, and Liner before joining academia.

The Data Analytics & Machine Intelligence (DAMI) Lab at Kangwon National University focuses on large-scale AI research, with expertise in natural language processing, multimodal AI, and knowledge extraction.

The lab focuses on:

• Natural Language Processing and Large Language Models (LLMs): Researching LLM fine-tuning, Chain of Thought (CoT), Retrieval-Augmented Generation (RAG), and AI Alignment. DAMI Lab also explores LLM Agent performance optimization and simulation-based learning.
• Machine Learning and Deep Learning: Addressing domain-specific AI challenges by constructing training datasets and developing deep learning solutions for real-world applications.
• Data Mining and AI-Driven Insights: Applying machine learning and LLM-powered analytics to fields such as social sciences and medical AI, uncovering insights to tackle complex problems.

Advancing AI research with Theta EdgeCloud

DAMI Lab will leverage Theta EdgeCloud, the first hybrid cloud-edge GPU computing platform, to drive innovation in natural language processing (NLP), machine learning, and data mining. Through this collaboration, DAMI Lab researchers will also gain access to Theta’s decentralized GPU infrastructure, providing cost-effective, scalable AI computing power to advance LLM training, multi-agent AI systems, and multimodal learning applications.

• Enhance NLP and LLM models: Train advanced language models with scalable, high-performance GPU computing.
• Develop robust multi-agent AI systems: Experiment with multi-agent debate models and adversarial AI.
• Power AI at the edge: Deploy AI models in real-time environments using Theta’s hybrid cloud-edge architecture.
• Optimize AI training efficiency: Leverage Theta’s distributed computing framework to run cost-efficient, large-scale AI experiments.

Addressing these developments, Professor Kyeongpil Kang states, “We’ve been evaluating EdgeCloud for a number of weeks already, and we’re deeply impressed by its flexibility and scalability. This will serve us well as our students, researchers and staff continue to grow over the next few years. We’re excited to leverage EdgeCloud’s distributed GPU architecture and see how DAMI Lab can accelerate NLP and multimodal AI research while optimizing model performance for real-world applications.”

Mitch Liu, co-founder and CEO of Theta Labs adds, “Bringing Kangwon National University onboard as our 23rd customer is exciting but also adds to our growth challenges. By pushing over 85% utilization rate across our GPU network, we need to constantly expand our GPU infrastructure while maximizing cost effectiveness. Our goal in 2025 is to 10x our customer base, drive over 95% utilization rate and lock over 2 billion TFUEL through our elite edge node operators.”