한국 Inception 스타트업을 위한 컨퍼런스 세션

Don't miss this keynote from NVIDIA founder and CEO Jensen Huang. He'll share how NVIDIA's accelerated computing platform is driving the next wave in AI, digital twins, cloud technologies, and sustainable computing.

Multi-modal, multi-agent AI systems are becoming the next-generation norm, and NVIDIA NIMs envisions the path to achieve it using optimized container templates. We'll explain how our GPU-native container engine further accelerates NIMs to deliver such advanced multi-agent AI systems at low cost and high performance. It exploits a novel fractional GPU-sharing technology to accommodate multiple different models having diverse performance bottlenecks in a single GPU, automate resource allocation and model combinations with memory size estimation techniques, and auto-scale the NIM containers by incorporating inference runtime metrics. All these features are implemented on both air-gapped on-premises clusters and cloud-native setups. On top of it, we've also built a streamlined UI to import, fine-tune, and serve open models in just one click, effectively hiding all those technical details from the end user. Putting it all together, let's dive into the universal world of NIMs.

Pre-training a non-English-centric large language model from scratch can be far more efficient than traditional approaches might suggest. By leveraging the vast pool of English data — while carefully injecting a small, high-quality corpus from the target language — we trained state-of-the-art multilingual models without incurring massive computational overhead. We'll introduce the concept of Knowledge Bridging, where a model’s “world knowledge” is largely learned through English, and then “bridged” to the target language with comparatively modest, region-specific data. See how this approach contrasts with naïve pre-training recipes (e.g., equal-ratio or monolingual emphasis) and why continual pre-training on an existing open-source checkpoint underperforms. We'll also highlight best practices for tokenizer design and data mixture strategies. Our real-world experiments demonstrate that this method not only sustains high performance on English benchmarks, but also significantly boosts target-language proficiency — making it an ideal recipe for building sovereign AI solutions in regions like Korea, Japan, and beyond. We'll close with the business implications of delivering custom LLMs to enterprises requiring robust, region-specific artificial intelligence.

The Korean AI Expert Day is an opportunity to meet Korean NVIDIA experts and connect with conference attendees from Korea. Exchange knowledge and industry trends and share GTC experiences. You can also network with Korean AI experts, creating international business or academic collaboration opportunities. Korean will be the predominant language spoken here.

Korean AI Expert Day는 한국 NVIDIA 전문가들을 만나고 한국에서 온 컨퍼런스 참석자들과 교류할 수 있는 기회입니다. 이 자리에서 지식과 업계 동향을 공유하고, GTC에서의 경험을 나눌 수 있습니다. 또한, 국제 비즈니스 및 학술 협력 기회를 창출하는 한국 AI 전문가들과의 네트워킹도 가능합니다. 본 행사에서는 한국어가 주요 언어로 사용됩니다.

We'll show how Samsung has accelerated its optical proximity correction (OPC) workloads using NVIDIA GPUs and the newly adopted NVIDIA cuLitho. With increasing lithography complexity in semiconductor manufacturing, GPU acceleration through cuLitho has enabled Samsung to achieve significant performance improvements, enhancing both speed and efficiency in critical OPC processes.

For most AI startups, "build it and they will come" doesn't describe what it takes to acquire customers. In a world of infinite alternatives, it takes ingenuity, grit, and committed partners to generate awareness and conversion for your product.

Join the Inception Sustainable Futures Showcase to discover how startups are harnessing NVIDIA technology for sustainable innovation. See real-world applications in climate entrepreneurship across high-impact domains such as agriculture, carbon capture, clean energy, climate and weather, environmental analysis, green computing, sustainable infrastructure, and waste management. You’ll also learn about the transformative impact of NVIDIA’s Earth-2 initiative.

We're bringing together visionary leaders from today's hottest startups to share insights shaping the future of business. Get an inside look at how these innovators are leveraging cutting-edge technology, bold strategies, and fresh perspectives to redefine industries and create disruptive growth. From lessons in leadership to the latest trends in innovation, this session offers invaluable takeaways for established executives and aspiring leaders alike. Discover what it takes to thrive in an era of rapid transformation and emerging opportunities.

Despite the persistent narrative that a lack of women-led companies explains why only 2% of venture capital funding goes to women, in reality female entrepreneurs are leading some of the most innovative ventures across various industries. This panel will challenge the myth of scarcity by spotlighting the growing number of women-led companies and the groundbreaking innovations they're driving. We'll explore the untapped potential of women-led businesses, highlight examples of cutting-edge technologies, and provide insights on how investors can capitalize on this thriving but overlooked sector. Join us as we reframe the conversation around investing in women and the transformative impact of their work.

Generative AI and foundational models are transforming healthcare and life sciences by enhancing efficiency, driving breakthroughs, and enabling tailored approaches to care. These technologies are creating a future where innovation changes how care is given, making patient outcomes and experiences much better. They're doing this by accelerating drug discovery, improving clinical workflows, and improving patient care. This panel brings together leaders from CytoReason, AITEM, Therapyside, and Superluminal Medicines, showcasing how NVIDIA inference microservices (NIMs), AI blueprints, and BioNeMo are driving advances in diagnostics, therapy workflows, and drug development. Through a technical and forward-looking discussion, we’ll explore how these innovators are integrating AI into their systems, overcoming challenges, and shaping a future where technology fundamentally transforms patient care and advances the potential for better health for all.

This session will feature innovative startups sharing their experiences and strategies for expanding beyond their local markets. Attendees will gain insights into the challenges and opportunities faced by these startups as they navigate global expansion.

Therapeutic Antibodies have six complementary determining regions (CDRs). CDRs are flexible loops with unique binding characteristics compared to other protein interactions. Between the CDRs, there are framework regions (FWRs) that support CDRs. Our research generates novel and antigen-specific antibodies by orchestrating CDR and FWR sequences. Additionally, we control the sequence similarity of generated antibodies to human antibody sequences to minimize undesired immune reactions. To do this, we use reward models that evaluate antibody properties. Reward models are improved by the wet lab assays using active learning. Also, we adopt antigen-antibody cross attention with chemical features to maximize the utilization of a small amount of data. Finally, three different antigens were tested through in vitro wet lab experiments to identify specific antibodies.

See how Linq-Embed-Finance improves the accuracy of financial data retrieval by up to 31% compared to general-purpose models. Leveraging the retrieval augmented generation (RAG) pipeline, Linq-Embed-Finance handles the complexities of financial documents such as earnings reports, 10-K/Q filings, and multistep queries. Learn how this model provides financial analysts with precise, context-rich information to make informed decisions across vast financial datasets.