- Took 1st place in Track C and Grand Prize among all 20 competing teams with synthetic data generation technology specialized for MoE quantization
- Built a dataset using an agent based on Nemotron 3 Super120B, presenting a data-centric rather than algorithm-centric optimization approach
SEOUL, South Korea, April 24, 2026 /PRNewswire/ — Nota AI, a leading AI model compression and optimization company, today announced that it took 1st place in Track C at the NVIDIA Nemotron Hackathon with its synthetic data generation technology specialized for Mixture of Experts (MoE) quantization, and was named Grand Prize winner among all 20 competing teams.
The Nota AI team that won the Grand Prize at the NVIDIA Nemotron Hackathon.
The hackathon was organized to share the latest research on NVIDIA’s newest open-source AI model, “Nemotron™”, and to give Korean developers hands-on experience with world-class AI technologies to strengthen their practical application capabilities. Participants competed across three tracks using NVIDIA technology: developing AI agents to solve real-world problems (Track A), advancing domain-specific Nemotron models through SFT and RL (Track B), and designing synthetic data pipelines to build high-quality datasets (Track C). Nota AI took 1st place in Track C, which focused on synthetic data pipelines for SDG, and received the highest overall evaluation among all competing teams to claim the Grand Prize.
In the competition, Nota AI used an agent based on NVIDIA’s Nemotron 3 super to build a quantization dataset specialized for the MoE architecture, and demonstrated a data-centric approach that minimizes performance loss in MoE models. While conventional quantization methods have primarily focused on formula- and algorithm-centric optimization, Nota AI differentiated itself by precisely engineering the structure, quality, and task-alignment of the dataset to unlock higher quantization performance.
This is also notable in that it demonstrates that the center of gravity in AI optimization is expanding beyond model compression algorithms themselves to encompass how data is designed and leveraged. Through this hackathon, Nota AI delivered tangible results aligned with this trend leveraging NVIDIA technology, and the achievement is expected to serve as a stepping stone to further expanding the two companies’ partnership. Nota AI is also continuing its technology collaboration with NVIDIA across vision AI applications by integrating NVIDIA’s video search and summarization tool “VSS Blueprint” into its vision-language model (VLM)-based real-time video analytics solution NVA (Nota Vision Agent), enabling real-time detection and summarization of anomalous situations and thereby reducing on-site response time.
Myungsu Chae, CEO of Nota AI, said, “This award demonstrates that AI optimization is not confined to algorithmic refinement alone — it shows that new possibilities can emerge depending on how we design and leverage purpose-fit data. Building on our partnership with NVIDIA, we will continue to advance data-centric AI optimization technologies and solutions that can be deployed directly to real-world industrial environments.”
