Step-by-Step Teaching Guide

---

STEP 1: Why AI Has Become a National Priority

The Big Picture

Governments are no longer treating AI as experimental. It has become national strategy.

Key Concept: From Pilot to Production

Why This Matters

• Nations want AI that reflects their local languages, cultures, and economic needs
• The goal is trusted, sovereign AI — not just imported solutions
• NVIDIA's stated mission: "Enable all nations to be AI nations"

✅ Check Your Understanding

Why would a government prefer locally controlled AI over a generic global solution?

---

STEP 2: NVIDIA's Core Tools — What They Are and What They Do

The Key Technologies Mentioned

🔷 NVIDIA Nemotron (Open Models)

• A family of open AI models that organizations can customize
• Supports text, vision, audio, and language (multimodal)
• Includes persona datasets — synthetic data reflecting local populations

🔷 NVIDIA NeMo Framework

• A developer toolkit for fine-tuning and training language models
• Used to adapt Nemotron to local languages (Thai, Vietnamese, etc.)

🔷 NVIDIA Cosmos

• Open world models for physical/real-world AI applications
• Used in smart city solutions (traffic, airports)

🔷 NVIDIA VSS Blueprint (Video Search & Summarization)

• Enables AI to search and summarize video content
• Applied in security, traffic management, and airport operations

🔷 NVIDIA cuOpt

• Software for optimization problems (routing, scheduling, logistics)

🔷 NVIDIA Vera Rubin Platform

• Advanced computing infrastructure for large-scale AI deployment
• The newest generation platform for enterprise AI

✅ Check Your Understanding

Match each tool to its primary use case: Nemotron → ?, NeMo → ?, Cosmos → ?, VSS → ?

---

STEP 3: Four Strategic Focus Areas for Public Sector AI

NVIDIA and partners are organizing their Southeast Asia work around four pillars:

┌─────────────────────────────────────────────────────┐
│  1. Government Operations & Service Delivery         │
│     → Smarter, faster public administration          │
├─────────────────────────────────────────────────────┤
│  2. Citizen Services & Local Business Empowerment    │
│     → Accessible AI for everyday people              │
├─────────────────────────────────────────────────────┤
│  3. Critical Infrastructure & Public Safety          │
│     → Security, traffic, emergency response          │
├─────────────────────────────────────────────────────┤
│  4. Startups, Developers & Researchers               │
│     → Building national AI innovation capacity       │
└─────────────────────────────────────────────────────┘

✅ Check Your Understanding

Which pillar does a legal AI chatbot for citizens fall under? What about a traffic monitoring system?

---

STEP 4: Country-by-Country Case Studies

Learning Through Real Examples

🇸🇬 Singapore

🇲🇾 Malaysia

🇹🇭 Thailand

🇻🇳 Vietnam

🇧🇳 Brunei

✅ Check Your Understanding

What do Viettel AI (Vietnam) and iApp Technology (Thailand) have in common in their approach?

(Answer: Both fine-tuned Nemotron models for their local language and applied them to legal AI use cases)

---

STEP 5: Understanding "Open Models" and Why They Matter

What Does "Open Model" Mean?

An open model is one where the weights, architecture, or training recipes are publicly available — allowing organizations to:

• Customize the model for their language or domain
• Control where data is processed (data sovereignty)
• Own the resulting model rather than depending on a vendor

The Nemotron Advantage for Southeast Asia

Generic Global AI Model
        ↓
Fine-tuned with local data (NeMo Framework)
        ↓
Region-Specific AI Model
(Thai legal language, Vietnamese culture, etc.)
        ↓
Deployed locally with data governance

Real Example: Thanoy (Thailand)

1. Base model: Nemotron 3 Nano
2. Fine-tuned with: Thai legal data
3. Result: OpenThai 2.0 Legal → released as open source
4. Application: Thanoy chatbot → 43,000 users served

✅ Check Your Understanding

Why would a government agency prefer an open model over a proprietary closed API?

---

STEP 6: Sea Limited — Enterprise-Scale AI Adoption

Who Is Sea Limited?

A Singapore-founded global tech company operating three major platforms:

Why Vera Rubin Matters Here

Sea is the first ASEAN enterprise to adopt NVIDIA Vera Rubin — the most advanced computing platform for AI.

This enables Sea to:

• Scale AI models to serve hundreds of millions of users
• Deploy intelligent agents across all three business lines
• Bring enterprise-grade AI to small businesses and underserved communities

Social Impact Angle

Small businesses on Shopee, previously unable to afford enterprise AI tools, now benefit from the same technology as large corporations.

Consumers underserved by traditional banks can access financial services through Monee's AI-powered credit assessment.

✅ Check Your Understanding

How does Sea's adoption of Vera Rubin create social and economic value beyond just business efficiency?

---

STEP 7: Connecting the Concepts — The Big Framework

How Everything Fits Together

NVIDIA Infrastructure (Vera Rubin)
            ↓
Open Models (Nemotron family)
            ↓
Developer Tools (NeMo, cuOpt, VSS, Cosmos)
            ↓
Local Fine-Tuning (language, culture, domain)
            ↓
Regional Applications
(Legal AI, Healthcare, Smart Cities, Finance, Gaming)
            ↓
Impact at Scale
(Citizens, SMEs, Governments, Researchers)

Three Recurring Themes

---

STEP 8: Final Synthesis — Key Takeaways

What You Should Now Know

✅ NVIDIA's role is providing the full stack: infrastructure → models → tools → ecosystem

✅ Nemotron open models are the foundation for Southeast Asian AI customization

✅ Fine-tuning allows generic models to become locally relevant and culturally appropriate

✅ Four strategic pillars guide public sector AI deployment across the region

✅ Real deployments exist today in Singapore, Malaysia, Thailand, Vietnam, and Brunei

✅ Sea Limited's Vera Rubin adoption demonstrates enterprise-scale AI serving social good

✅ Open source releases (like OpenThai 2.0 Legal) strengthen national AI ecosystems

---

🧪 Final Assessment Questions

1. Explain why a country like Thailand would fine-tune an existing model rather than build one from scratch.

2. Compare how Nemotron is being used differently in Vietnam versus Brunei.

3. Analyze how Sea Limited's AI applications across Shopee, Monee, and Garena each address a different social need.

4. Evaluate the importance of releasing AI models as open source for national AI development.

5. Describe the journey from a base Nemotron model to a deployed citizen-facing application, using any country example from the article.