Event 2

Summary Report: “Cloud Mastery (#1 AI From Scratch)”

Event Objectives

  • Introduce Strands Agents: enabling developers to build agents with simpler and more concise code
  • Present methods for effective LLM prompting (the art of communicating with AI) to achieve high-quality outputs
  • Share a case study of an AIoT Project: Locker Manager (IoT & AWS services)

Speakers

  • Banh Cam Vinh
  • Nguyen Tuan Thinh – DevOps Engineer, First Cloud AI Journey
  • Aiden Dinh – Katalon Operation Engineer

Key Highlights

Limitations of Traditional Application Architecture

  • Code for integrating tools/agents is often long and complex to configure → time-consuming and error-prone
  • Poor prompting skills → responses do not meet expectations, requiring multiple retries → wastes time and tokens
  • Manual management of borrowing learning tools → inconvenient when staff are absent, difficult to control

Transitioning to modern application architecture

1. Strands Agents

  • Open-source solution that enables building agents with fewer lines of code and simpler configurations
  • Benefits: ease of use, native tools, MCP server support, seamless AWS integration, scalability, and rapid development
  • Case Study – Customer Support Agent: Demonstrates how to quickly build an agent using Amazon Bedrock, define and attach tools (tasks), and design workflows

2. Enhancing LLM Output Quality

  • The art of communicating with AI: identifying common prompting issues and introducing key components of effective prompts to optimize both output quality and token cost
  • Case Study – Protimizer: A Chrome browser extension that helps optimize prompts and allows continuous AI interaction across multiple tabs without losing conversation context
  • Highlighted AWS services: Amazon Cognito, Amazon Bedrock, AWS Lambda, Amazon DynamoDB

3. Case Study – AIoT Project

  • Mission: Design an automated locker management system
  • Technique: Use Arduino and sensors to monitor items, combined with AWS IoT Core and Amazon Rekognition for device connectivity and facial recognition
  • Highlighted AWS services:
    • AWS IoT Core (acts as a secure intermediary enabling communication between IoT devices and AWS services)
    • Amazon Rekognition (performs facial recognition by comparing captured images with stored member images in an S3 bucket)

Key Takeaways

  • Strands Agents: Simplify development, improve scalability, and integrate easily with AWS services compared to traditional architectures (which are complex, less scalable, and time-consuming).
  • Prompt Engineering Mindset: Learn how to design effective prompts to optimize output quality and reduce LLM usage costs.
  • Case Study Insights: Understand how AWS services can be applied to build systems that improve performance and enable automation.

Event Experience

Participating in the “Cloud Mastery” workshop was a highly valuable experience that gave me a comprehensive view of modernizing applications and databases using advanced methods, services, and tools. Through real-world case studies, I gained a clearer understanding of how to deploy systems on the cloud and better visualize AWS infrastructure architecture.

Networking & Knowledge Sharing

  • The workshop provided opportunities to interact directly with speakers, experts, and the tech team, helping me expand my network and gain valuable insights from the community.

Some event photos

Overall, the event not only provided technical knowledge but also helped reshape my mindset in application design and how to effectively apply what I have learned to real-world projects.