Skip to Content Facebook Feature Image

FPT and NVIDIA Expand Nemotron-Personas-Vietnam Dataset by 10X Times to Power the Next Wave of Agentic AI

Business

FPT and NVIDIA Expand Nemotron-Personas-Vietnam Dataset by 10X Times to Power the Next Wave of Agentic AI
Business

Business

FPT and NVIDIA Expand Nemotron-Personas-Vietnam Dataset by 10X Times to Power the Next Wave of Agentic AI

2026-10-16 17:18 Last Updated At:18:25

HANOI, Vietnam, Oct. 16, 2026 /PRNewswire/ -- FPT announced the expansion of Nemotron-Personas-Vietnam, its open dataset of synthetic personas grounded in Vietnam's official demographic and labor statistics. The dataset now contains 1,000,000 personas, giving developers, researchers, and enterprises a larger and more representative foundation for building models, agents, and applications that understand Vietnamese people, language, and culture.

At a Glance

First released by FPT in collaboration with NVIDIA in June 2026, Nemotron-Personas-Vietnam is an open dataset of synthetic personas grounded in Vietnam's official statistics from the General Statistics Office (GSO), including the Population and Housing Census 2024 and the Vietnam Household Living Standards Survey (VHLSS) 2024, as well as from FPT Smart Cloud and Quantum AI & Cyber Security Institute (FPT Corporation).

  • 21 fields: 6 persona fields and 15 contextual fields covering persona attributes and demographics.
  • 13,000 unique full-name combinations.
  • 20 occupational categories derived from labor survey data.

Built on NVIDIA's Nemotron-Personas methodology, which anchors persona generation in real-world statistics and uses open-weight LLMs to write high-fidelity persona narratives, the dataset gives developers realistic Vietnamese user profiles spanning age, region, occupation, income, education, and cultural context, all without any personally identifiable information (PII).

A Dataset That Understands Vietnam at Scale

This expansion is more than a larger dataset. Growing to 1,000,000 personas opens new ways for developers to build, train, and evaluate AI for Vietnam:

  • Finer segmentation: Users can sample highly specific slices of the Vietnamese population, such as personas that represent a retiree in Ho Chi Minh City, a smallholder farmer in the Mekong Delta, or a young engineer in Da Nang, with enough records in each to train and test reliably.
  • Larger workloads: The expanded scale supports bigger training runs, more comprehensive red-teaming, and population-level simulations that the original release could not sustain on its own.
  • Reduced model collapse: Richer variation in narratives, skills, and behaviors lowers the risk that models trained on the data collapse onto a narrow set of templates, a growing concern as synthetic data becomes load-bearing in AI pipelines.
  • Deeper coverage: the earlier release sampled a set of major provinces and cities. The 1,000,000-persona version covers the whole country, so developers can build and test for users in every region.

The dataset remains fully synthetic, contains zero personally identifiable information, and is released openly under a permissive license on Hugging Face, compatible with NVIDIA NeMo libraries.

A Commitment to Vietnam's Sovereign AI

An expansion sends a different signal than a launch. It tells developers that the resource they started building on is being maintained, grown, and invested in, not left as a one-time release.

For FPT, this is a deliberate statement of long-term commitment. Vietnam is accelerating its digital transformation. Under the newly approved Program for Digital Economy and Digital Society Development 2026-2030, the government targets the digital economy to reach approximately 30% of GDP by 2030. Across banking, finance, healthcare, insurance, retail, and public services, organizations increasingly need realistic persona data for product testing, service personalization, simulation, and AI model evaluation. By growing the dataset rather than leaving it as-is, FPT is building the data layer of Vietnam's sovereign AI, AI grounded in the country's own population, language, and values rather than imported from elsewhere.

Fostering Agentic AI Capabilities in Practice

As agentic AI emerges as the next frontier, systems that reason, act, and serve specific users rather than just answer prompts, the demand for grounded, identity-aware data will only grow. A larger, more representative persona dataset gives the Vietnamese AI community the foundation to build agents that genuinely understand who they are serving.  

Customer Personas: Build customer-facing agents that can draw on personas spanning income brackets, regions, and financial-literacy levels to adjust their  tone, product recommendations, and compliance handling for each customer, such as a retiree in Hanoi versus a young professional in Ho Chi Minh City.

Simulation and Evaluation for Agentic Systems: Create simulation environments populated with realistic Vietnamese users and organizations, run red-teaming and bias-and-fairness evaluations across diverse demographic groups, and stress-test how agents behave across regions, ages, and occupations, all without touching real customer data.

Domain-Specific AI Agents: Enable specialized AI agents across regulated industries, including finance, healthcare, insurance, and public services, that need to reason about who they are helping. A healthcare agent can be evaluated against patient personas spanning literacy levels, regions, and care settings. A public-service agent can be tested against citizen-personas across the country's administrative landscape.

Download the dataset for free on Hugging Face.

** This press release is distributed by PR Newswire through automated distribution system, for which the client assumes full responsibility. **

FPT and NVIDIA Expand Nemotron-Personas-Vietnam Dataset by 10X Times to Power the Next Wave of Agentic AI

FPT and NVIDIA Expand Nemotron-Personas-Vietnam Dataset by 10X Times to Power the Next Wave of Agentic AI

Collaboration supports EU CRA readiness and cybersecurity planning for industrial and edge AI applications.

TAIPEI, Oct. 16, 2026 /PRNewswire/ -- Advantech, a global leader in IoT intelligent systems and embedded platforms, has signed a memorandum of understanding (MOU) with DEKRA, a global testing, inspection and certification organization, to strengthen cooperation on industrial cybersecurity compliance.

As the European Union's Cyber Resilience Act (CRA) is implemented in stages, manufacturers of industrial equipment and edge AI products need to address cybersecurity requirements earlier in product development. The partnership combines Advantech's industrial computing and application experience with DEKRA's cybersecurity testing and international certification expertise. It aims to help end customers and system integrators plan for EU CRA requirements and international standards such as IEC 62443, reduce compliance risks and avoid delays caused by late-stage changes.

Stan Zurkiewicz, CEO of DEKRA, said: "Digital trust is essential for industrial innovation to succeed in global markets. As AI and connected technologies become integral to manufacturing, energy and transportation, companies need verifiable evidence of safety and cybersecurity to build confidence in new technologies. Our collaboration with Advantech combines its industry experience with DEKRA's independent third-party testing and certification expertise. DEKRA's role is to provide independent assurance, helping companies navigate evolving international requirements and build trust in new technologies."

Miller Chang, President of Advantech's Embedded Sector, said at the signing ceremony: "As edge AI advances, customers increasingly need to consider compliance from the outset of product design. By combining Advantech's industrial computing and edge AI expertise with DEKRA's global testing and certification expertise, we aim to help customers identify regulatory requirements earlier, reduce compliance risks and shorten time to market, bringing trustworthy edge AI innovations to global markets more efficiently."

Planning for Cybersecurity Compliance from the Design Stage

The collaboration aims to help manufacturers consider cybersecurity and compliance requirements at an early stage of product development, drawing on DEKRA's independent testing and certification expertise. It will help customers identify regulatory and technical requirements across markets, plan assessment and certification pathways, and address potential cybersecurity gaps during development.

The scope includes independent technical assessments, documentation reviews, certification readiness assessments and cybersecurity testing to help manufacturers identify and address potential compliance gaps. Early planning can help reduce the time and costs associated with late-stage design changes and establish a clear compliance process from product concept and development through testing and market deployment.

First Milestone: AFE-E630 Security Functions Meet IEC 62443-4-2 SL2 Requirements

As an initial outcome of the collaboration, Advantech's AFE-E630 industrial edge system has passed an independent assessment by DEKRA and received an Attestation of Conformity (AoC). Designed for commercial and industrial energy storage management applications, the AFE-E630 has security functions that meet the relevant Security Level 2 (SL2) requirements of IEC 62443-4-2:2019 within the assessed scope. The assessment results provide a verified basis for product cybersecurity in these applications and a reference for customers' subsequent product integration and system-level compliance assessments.

Exploring Further Applications in Edge AI and Industry

Beyond existing industrial applications, Advantech and DEKRA plan to extend their cooperation to AI-enabled industrial systems, semiconductor manufacturing, robotics, drones and other safety-critical industrial applications, as well as energy and transportation applications.

As edge AI is integrated into physical equipment, product safety, functional safety and cybersecurity need to be considered together. The companies will combine Advantech's edge computing platforms and application experience with DEKRA's testing, cybersecurity and functional safety expertise to support the safety and trustworthiness of next-generation intelligent devices.

Through the MOU, Advantech and DEKRA will continue to develop their technical cooperation, helping customers address market-specific compliance requirements and prepare industrial and edge AI products for international deployment.

About DEKRA

For more than 100 years, DEKRA has been a trusted name in safety. Founded in 1925 with the original goal of improving road safety through vehicle inspections, DEKRA has grown to become the world's largest independent, non-listed expert organization in the field of testing, inspection, and certification. Today, as a global partner, the company supports its customers with comprehensive services and solutions to drive safety and sustainability forward. In 2025, DEKRA generated revenue of 4.4 billion euros. More than 48,000 employees are providing qualified and independent expert services in approximately 60 countries across five continents. DEKRA holds a Platinum sustainability rating from EcoVadis, placing it among the top 1% of the globally assessed companies.

** This press release is distributed by PR Newswire through automated distribution system, for which the client assumes full responsibility. **

Advantech and DEKRA Sign MOU to Advance Industrial Cybersecurity Compliance

Advantech and DEKRA Sign MOU to Advance Industrial Cybersecurity Compliance

Advantech and DEKRA Sign MOU to Advance Industrial Cybersecurity Compliance

Advantech and DEKRA Sign MOU to Advance Industrial Cybersecurity Compliance

Recommended Articles