AWS First Cloud AI Journey - Community Day

Summary Report: “AWS First Cloud AI Journey – Community Day”

Event Objectives

• Connect the FCAJ (First Cloud AI Journey) community in Ho Chi Minh City
• Share practical experience working with AI / GenAI in real production workloads
• Update useful AWS services for AI workloads (CloudFront, Amazon Quick suite)
• Provide career inspiration through hackathon stories and enterprise case studies

Speakers

• Tinh Truong – Platform Engineer at GoTymeX
• Thinh Nguyen – DevOps Engineer | FCAJ Ambassador
• Anh Pham – Cloud Consultant at G-AsiaPacific Vietnam
• Uyen Le – GenAI Engineer at VIB
• Thao Nguyen – GenAI Engineer at VIB
• Mai Nguyen – GenAI Engineer at VIB
• Duc Dao – Solutions Architect at Cloud Kinetics
• Vy Lam – Senior Business Systems Analyst at VPBank

Agenda Timeline

Key Highlights

Tinh Truong – Context Is Everything: Making AI Actually Work for You

• Why AI fails without context, and what “context” really means
• From prompts to memory: how AI is evolving with the Second AI Brain concept
• How better context produces better results — practical mindset and tips
• Career insights and how students can start building with AI

Anh Pham – Friendly AI Assistant with Amazon Quick

• Quick Chat Agent: AI assistants for exploring and analyzing data
• Quick Flows: build intelligent workflows with natural language, no coding required
• Quick Spaces: collaborative spaces that turn individual insights into team knowledge
• Quick Sight: build dashboards and reports from raw data using natural language

Thinh Nguyen – From Edge To Origin: CloudFront as Your Foundation

• Amazon CloudFront for every workload type
• Cost optimization techniques with CloudFront
• Security capabilities at the edge
• Enhanced reliability and performance with CloudFront

Uyen Le, Thao Nguyen, Mai Nguyen – 36 hrs with LotusHacks: Building UTMorpho

• Why the team joined LotusHacks
• From zero to idea — the brainstorming journey
• Defining the problem and shaping UTMorpho
• Building under pressure during a 36-hour development sprint
• Challenges, failures, and turning points
• Product overview, demo, and key learnings

Duc Dao – Non-Determinism of “Deterministic” LLM Settings

• How LLMs choose the next token
• The common assumption that Temperature = 0 guarantees determinism
• Reality: inference optimizations break that assumption
• Practical impacts on production systems and mitigation strategies

Vy Lam – Enterprise-Grade Multi-Agent System: Startup Credit Scoring

• The structural mismatch between banking systems and startup data
• Single Agent: when to use it and when not to
• The Multi-Agent paradigm for complex domains
• Blueprint of a Virtual Credit Committee
• Guardrails, compliance, operational ROI, and an implementation roadmap

Key Takeaways

Mindset for working with AI

• Context engineering matters more than clever prompting
• Treat AI like a teammate that needs memory, tools, and grounding, not a search engine
• Determinism is an illusion — always design for variance, even with temperature=0

Architecture & AWS

• CloudFront is more than a CDN; it can be a security and performance foundation
• Amazon Quick suite lowers the barrier for non-engineers to build AI workflows
• For complex business domains, multi-agent systems with guardrails outperform a single agent

Career & community

• Hackathons like LotusHacks are a fast way to ship something real and learn under pressure
• Enterprise AI adoption requires compliance, ROI tracking, and a clear roadmap, not just demos
• Joining the FCAJ community is a shortcut to learning from people already doing this in production

Applying to Work

• Add a context layer (memory, retrieval, tools) to internal AI assistants instead of relying on raw prompts
• Pilot Amazon Quick Flows / Quick Sight for internal reporting use cases
• Audit current LLM features for non-determinism risks and add guardrails or evaluation harnesses
• Evaluate CloudFront for static asset delivery and basic edge security on new projects
• Sketch a multi-agent design for any workflow that currently struggles as a single-prompt pipeline

Event Experience

Attending AWS First Cloud AI Journey – Community Day on May 23, 2026 was an inspiring half-day filled with both deep technical content and real-world stories. Key moments included:

Learning from a diverse speaker lineup

• Speakers came from very different angles — banking (VPBank, VIB), platform engineering (GoTymeX), AWS partners (G-AsiaPacific, Cloud Kinetics) and the FCAJ community itself.
• Each talk hit a different layer of the AI stack: prompting, tooling, infrastructure, and enterprise architecture.

Strong technical takeaways

• Tinh Truong reframed how I think about prompting — context first, prompt second.
• Duc Dao broke the comfortable myth that temperature=0 means reproducible output.
• Vy Lam showed how multi-agent systems are already being used in serious financial domains, not just demos.

Hands-on AWS exposure

• Anh Pham demoed the Amazon Quick suite, which lowers the bar for building internal AI tools.
• Thinh Nguyen went deep on CloudFront beyond CDN — security, cost, and reliability angles I had underused.

Inspiration from peers

• The UTMorpho / LotusHacks story from Uyen Le, Thao Nguyen, Mai Nguyen was a strong reminder that 36 hours is enough to ship something meaningful with the right team.

Networking

• Met other FCAJ/FCJ members at floor 26 of Bitexco, exchanged notes on internships, AWS learning paths, and current AI projects.

Some event photos

Overall, the event reinforced that building useful AI products is mostly about context, guardrails, and team execution — and that the FCAJ community in HCMC is a great place to learn this together.