Event 2

Summary Report: “AWS Cloud Day Vietnam 2025 – AI Edition”

Theme & Tagline: New Age Vietnam: From Vision to Value — turning AI vision into measurable value.
Focus: Generative AI with complementary Cloud capabilities (data, security, infrastructure, modern apps).

Event Objectives

  • Drive Vision → Value: tie GenAI to clear KPIs/ROI (speed, cost, quality).
  • Present GenAI reference architectures (RAG, agentic workflows, data pipelines, guardrails).
  • Define a practical adoption path: Idea → PoC → Pilot → Scale with governance & data safety.
  • Equip builders/engineers and IT leaders with AWS best practices (observability, cost control).

Speakers

  • AWS Vietnam leaders & specialists (AI/GenAI, Data, Security, AppMod).
  • Customer speakers from Vietnam (transformation & GenAI use cases).
  • Partners (deployment playbooks, cost optimization, operations).

Key Highlights

Central theme: Generative AI on AWS

  • Amazon Bedrock / Amazon Q & Guardrails: rapid start, security, compliance.
  • Data-first: retrieval/rerank/embeddings quality drives output quality.
  • Cloud consistency: build on AWS foundations (identity, networking, storage, monitoring) for security & scale.

Representative tracks

  • GenAI Foundations & Architecture: RAG, tool-use/agents, evaluation & observability.
  • Data, Security & Governance: PII protection, access control, moderation, cost/token tracking.
  • App Modernization for AI: service decomposition, API-first, event-driven; making apps AI-ready.
  • Builders Hands-on: demos/mini-labs for internal assistants and semantic search.

Key messages

  • Vision → Value: avoid demo-only efforts; anchor to KPIs from day one.
  • Data before models: optimize pipelines, indexing, caching, and latency.
  • Safety & compliance: guardrails, auditability, continuous monitoring & evaluation.

Key Takeaways

Design Mindset

  • Business-first: small, high-impact problems with clear owners & data.
  • Ubiquitous language: align business–tech and lock KPIs/ROI early.

Technical Architecture

  • Prefer RAG and agentic patterns; separate orchestrator from tools.
  • Observability & Eval: tracing, offline/online evaluation, feedback loops.
  • Cost & Performance: fit-for-purpose models, batching/caching, prompt/context optimization, latency control.

Modernization Strategy

  • Roadmap PoC → pilot → limited rollout → scale with exit criteria per stage.
  • Build on AWS foundations for security, resilience, and scalability.

Applying to Work

  • Shortlist 1–2 GenAI use cases tied to clear KPIs; prepare data & access.
  • Run a 6–8 week PoC with go/no-go gates (quality, latency, cost/interaction, adoption).
  • Ops plan: monitoring, guardrails, human-in-the-loop, continuous improvement.

Event Experience

The event offered a clear path to move from vision to value: deploying GenAI into real workflows with security, cost awareness, and business alignment. Talks and demos showed how to standardize the data pipeline, implement RAG/agents, and measure impact to unlock pilot & scale.