South Korea’s Electronics and Telecommunications Research Institute (ETRI) has released TANGO, an open-source, no-code platform designed to help people without specialized AI expertise build and deploy models through a browser-based workflow.

The platform brings data preparation, model training and optimization, and software deployment into one workflow, reducing reliance on separate data, AI, and software specialists.

Developed since 2021, TANGO 1 focused on computer vision, while TANGO 2 expands the platform to generative AI and large language models (LLMs), including training and management with industry-specific data.

TANGO 2 supports generative-AI development tasks such as data preprocessing and augmentation, fine-tuning, knowledge distillation, model optimization, and deployment.

Users can specify an application area and target device to receive suitable model recommendations, model conversions, or execution code tailored to that environment.

ETRI has released TANGO 2’s core LLMOps technology on GitHub, including microservices, Kubernetes-based infrastructure, web interfaces, shared libraries, and tools for data augmentation, model evaluation, application operation, and deployment.

The repository also includes code for field verification in areas such as software-defined ships, manufacturing, and factories.

Applications include AVENOTICS’ AI-supported autonomous ship navigation, Sure Data Lab’s agricultural knowledge service, and Seoul National University Hospital’s automated chest CT reporting technology.

Seoul National University Hospital and Seoul National University Bundang Hospital also plan to demonstrate a service that predicts cardiopulmonary and vascular disease.

ETRI is transferring the technology to industry: Acryl is incorporating TANGO LLMOps into its Jonathan product, while Persona AI received a technology transfer in April and is expanding LLMOps capabilities for its GEN STATION AI-agent platform.

ETRI senior researcher Cho Chang-sik said the tool is intended to help non-specialists build practical AI services.

ETRI is pursuing international standardization of no-code LLMOps technology through the ITU Telecommunication Standardization Sector, with final approval targeted for May 2027.