← Home
AI FOR THAIS
CPG-TDA National AI Innovation Hackathon with AWS

AWS Generative AI Training
& Hackathon Guide

Learn to build AI-powered applications using AWS PartyRock, Amazon Kiro, and Amazon Bedrock. Register, train, and submit your capstone project for the competition.

Target: Thai citizens 18+ | Timeline: Augโ€“Dec 2026

๐Ÿ“‹ Register for the Hackathon

Fill in your details below to register for the CPG-TDA AI Innovation Hackathon with AWS. All fields marked * are required.

Participant Registration

๐Ÿ“… Key Dates

Registration Opens26 August 2026
Registration Closes11 September 2026
Training Period14 Sep โ€“ 26 Oct 2026
Submission Deadline6 November 2026
Top 20 Bootcamp18โ€“20 November 2026
Awards Ceremony10 December 2026

๐ŸŽ‰ What is AWS PartyRock?

AWS PartyRock is a no-code generative AI app builder powered by Amazon Bedrock. It allows you to create AI applications without writing any code โ€” perfect for rapid ideation and prototyping.

Why PartyRock for this Hackathon?

  • No coding required โ€” anyone can build AI apps
  • Powered by foundation models โ€” Claude, Llama, Titan, and more
  • Instant deployment โ€” share your app with a URL
  • Free to use โ€” no AWS account needed
  • Perfect for ideation โ€” validate ideas quickly before building with Kiro/Bedrock
PartyRock is your starting point! Build your idea here first, then the Top 20 teams will enhance their prototype using Kiro and Bedrock.

๐Ÿš€ Getting Started

1
Go to PartyRock

Visit partyrock.aws and sign in with your Google, Apple, or Amazon account.

2
Explore Existing Apps

Browse the gallery to see what others have built. Get inspiration!

3
Click "Build your own app"

Describe what you want to build in plain English. PartyRock generates the app structure automatically.

๐Ÿ› ๏ธ Build Your First App

Let's build a simple disaster alert generator as an example:

Build me an app that takes a location in Thailand and generates a natural disaster risk assessment with recommended safety measures and emergency contacts.

PartyRock creates an app with:

  • Text input widget for location
  • AI-generated risk assessment
  • Safety recommendations panel
  • Emergency contacts output

Customize Your App

  • Click any widget to edit its prompt/settings
  • Add new widgets (text generation, image generation, chatbot)
  • Rearrange layout by dragging widgets
  • Change the AI model powering each widget
  • Click "Publish" to get a shareable URL

๐Ÿงฉ Understanding Widgets

Widget TypeWhat It DoesUse Case
User InputText field for user to typeLocation, topic, question
Text GenerationAI generates text based on promptAnalysis, recommendations, summaries
Image GenerationAI creates images from textVisualizations, diagrams, mockups
ChatbotInteractive conversation with AIQ&A assistant, advisor
Connect widgets together! Use the output of one widget as input to another to create powerful chains of AI processing.

๐ŸŽฏ Example Project Ideas

๐ŸŒŠ Flood Risk Analyzer

Input: Thai province โ†’ Output: flood risk level, historical data, evacuation recommendations, emergency contacts

๐ŸŒพ Crop Disease Identifier

Input: crop type + symptoms โ†’ Output: likely disease, treatment steps, prevention tips

๐Ÿฅ Health Symptom Checker

Input: symptoms in Thai โ†’ Output: possible conditions, urgency level, nearby clinics

๐Ÿ“š AI Study Planner

Input: subject + exam date โ†’ Output: personalized study schedule, key topics, practice questions

๐Ÿ™๏ธ Air Quality Advisor

Input: location + activity type โ†’ Output: air quality assessment, health recommendations, best times for outdoor activity

๐Ÿ’ก Tips for the Hackathon Submission

โœ… What Judges Look For

  • Problem relevance: Does it solve a real Thai challenge?
  • AI usage: Is the AI doing meaningful work (not just decoration)?
  • User experience: Is it intuitive and useful?
  • Creativity: Novel approach to the problem?
  • Presentation: Can you clearly explain your idea?

๐Ÿค– What is Amazon Kiro?

Amazon Kiro is an AI-powered development environment that writes code for you. Think of it as a programming partner that understands natural language and can build full applications from your descriptions.

What Kiro Can Do

  • Write complete applications from natural language instructions
  • Create files, folders, and project structures automatically
  • Run terminal commands (install, build, run, test)
  • Fix errors and debug code automatically
  • Explain technical concepts in simple terms
  • Iterate on your requirements step by step
You don't need to memorize code! Just describe what you want in English, and Kiro writes the code for you.

๐Ÿ› ๏ธ Setup & Installation

Requirements

  1. Download Kiro IDE โ€” Free at kiro.dev
  2. Node.js 18+ โ€” Download at nodejs.org
  3. Modern browser โ€” Chrome, Firefox, or Edge
  4. Internet connection โ€” Required for AI features
1
Install Kiro

Download from kiro.dev, run the installer, and sign in with your AWS Builder ID (free).

2
Create a New Workspace

Open Kiro โ†’ File โ†’ New Workspace โ†’ Choose a folder for your hackathon project.

3
Start Chatting with Kiro

Open the chat panel and describe your project. Kiro will ask clarifying questions and start building!

๐Ÿ“ Your First Project with Kiro

Let's build a disaster alert dashboard as an example for the hackathon:

I want to build a web application that shows real-time flood risk levels across Thai provinces. Users can select a province and see: risk level (low/medium/high/critical), historical flood data, recommended actions, and emergency contacts. Make it mobile-friendly with a modern UI.

Kiro will:

  • Ask clarifying questions about data sources and design preferences
  • Create the full project structure (15+ files)
  • Build the frontend UI with interactive map/dashboard
  • Add sample data for Thai provinces
  • Install dependencies and verify everything works
  • Run the app so you can see it in your browser

๐Ÿ’ก Pro Tips

  • Be specific: "Add a red banner for critical alerts" vs "make it look nice"
  • Iterate: Start simple, add features one at a time
  • Ask for explanations: "Explain how the alert system works"
  • Request changes: "Change the color scheme to blue and white"

๐Ÿงญ Kiro Steering

Steering files are rules that Kiro follows every time you interact with your project. Write them once, and Kiro always remembers โ€” no need to repeat yourself.

your-project/
โ”œโ”€โ”€ .kiro/
โ”‚   โ””โ”€โ”€ steering/
โ”‚       โ”œโ”€โ”€ project-context.md
โ”‚       โ”œโ”€โ”€ design-guide.md
โ”‚       โ””โ”€โ”€ coding-standards.md
โ”œโ”€โ”€ src/
โ””โ”€โ”€ package.json
Create steering files for my hackathon project. The app is a Thai disaster alert system. Use Next.js, Tailwind CSS, mobile-first design, and English/Thai bilingual UI. Primary color should be emergency-red (#dc2626) for alerts.

Kiro creates steering files that ensure every new feature you add automatically follows these rules โ€” consistent colors, bilingual text, mobile-responsive layout, and the correct tech stack.

๐Ÿ“‹ Spec-Driven Development

For complex features, use Kiro's Spec mode โ€” it breaks down your idea into requirements โ†’ design โ†’ implementation tasks, then builds each piece systematically.

When to Use Spec Mode

  • Building a complete feature with multiple components
  • Need clear documentation of what was built and why
  • Working in a team and need shared understanding
  • Want the AI to plan before coding
Use spec mode to build an emergency notification system that: 1) monitors flood sensor data, 2) triggers alerts when water levels exceed thresholds, 3) sends SMS-style notifications to registered users, 4) logs all alerts with timestamps.

โšก Hackathon Workflow with Kiro

1
Ideate โ€” Describe your hackathon project idea in the chat
2
Scaffold โ€” Let Kiro create the project structure and basic UI
3
Iterate โ€” Add features one by one with follow-up prompts
4
Polish โ€” Request responsive design, better UX, error handling
5
Present โ€” Ask Kiro to help prepare your demo pitch deck

๐Ÿ”๏ธ What is Amazon Bedrock?

Amazon Bedrock is a fully managed service that offers high-performing foundation models (FMs) from leading AI companies through a single API. It's the production-grade AI backbone for your application.

Key Capabilities

  • Text generation โ€” Write, summarize, translate, analyze
  • Image generation โ€” Create visuals from descriptions
  • Embeddings โ€” Convert text to vectors for search/similarity
  • Agents โ€” Build autonomous AI workflows
  • Knowledge Bases โ€” Ground AI responses in your own data
  • Guardrails โ€” Control AI behavior and safety
For this hackathon, the Top 5 finalists will use Bedrock to upgrade their PartyRock/Kiro prototypes into production-ready MVPs with real AI capabilities.

๐Ÿง  Available Foundation Models

ModelProviderBest For
Claude 3.5 SonnetAnthropicComplex reasoning, coding, analysis
Claude 3 HaikuAnthropicFast responses, chat, simple tasks
Llama 3MetaGeneral text, multilingual (Thai)
Titan TextAmazonSummarization, search, embeddings
Titan ImageAmazonImage generation & editing
Stable Diffusion XLStability AIHigh-quality image generation

๐Ÿ”‘ How to Access Bedrock

For Hackathon Participants (Top 5)

AWS will provide temporary access to Bedrock during the finalist development phase (23โ€“27 Nov). You'll receive:

  • AWS account credentials with Bedrock access
  • Spending limits pre-configured
  • Technical mentorship from AWS Solutions Architects
1
Sign in to AWS Console

Use the credentials provided by the hackathon organizers

2
Navigate to Bedrock

Search "Bedrock" in the AWS Console โ†’ Click "Amazon Bedrock"

3
Enable Model Access

Go to "Model access" โ†’ Request access to the models you need

4
Start Building

Use the Playground to test prompts, then integrate via API

๐ŸŽฎ Using the Bedrock Playground

The Playground lets you test different models and prompts before writing any code:

Playground Modes

  • Text: Single prompt โ†’ response (great for testing)
  • Chat: Multi-turn conversation (build chatbots)
  • Image: Generate images from text descriptions
You are a Thai disaster response expert. Given a flood alert for [province], provide: 1) Severity level, 2) Affected areas, 3) Immediate actions for residents, 4) Emergency contacts. Respond in both Thai and English.

Test your prompts in the Playground, then refine them before integrating into your application code. This saves time and API costs.

๐Ÿ”Œ Integrating Bedrock in Your App

Use the AWS SDK to call Bedrock from your application (Kiro can help you write this code):

Add Amazon Bedrock integration to my app. When a user enters a Thai province, call Claude on Bedrock to generate a flood risk assessment. Display the result in a nicely formatted card.

Kiro will:

  • Install the AWS SDK
  • Create an API route that calls Bedrock
  • Handle the response and format it for display
  • Add loading states and error handling

๐Ÿ† Bedrock for Top 5 Finalists

Your MVP Development Phase (23โ€“27 Nov)

During this 5-day sprint, you'll upgrade your prototype to a production-ready MVP using Bedrock:

  • Replace mock/sample data with real AI-generated insights
  • Add Knowledge Bases to ground responses in real Thai data
  • Implement Guardrails for safe, responsible AI outputs
  • Build Agents for autonomous multi-step workflows
  • Prepare a compelling demo for the Awards Ceremony

๐ŸŽฏ Competition Overview

The CPG-TDA AI Innovation Hackathon challenges you to build an AI-powered solution addressing Thailand's most pressing challenges. Your capstone project demonstrates your ability to apply Generative AI to real-world problems.

Theme

AI solutions for Thailand's pressing challenges. Focus areas include:

  • ๐ŸŒŠ Natural Disaster & Public Safety (Predict, Detect & Alert, Reach & Respond)
  • ๐ŸŒพ Smart Agriculture & Food Security
  • ๐Ÿฅ Healthcare & Public Wellness
  • ๐Ÿ“š Education & Workforce Development
  • ๐Ÿ™๏ธ Sustainable Cities & Environment

๐Ÿ—บ๏ธ Your Hackathon Journey

1
Training Phase (14 Sep โ€“ 26 Oct)

Complete Gen AI foundational training. Learn PartyRock, Kiro basics, and AI concepts.

2
Ideation & Submission (27 Oct โ€“ 6 Nov)

Build your capstone project on AWS PartyRock. Submit your app URL + project description.

3
AI Judging โ†’ Top 20 (9โ€“10 Nov)

AI-automated scoring selects Top 20 teams based on judging criteria.

4
Bootcamp (18โ€“20 Nov)

Top 20 teams attend 3-day bootcamp at CPG-TDA venue. Build prototypes with Amazon Kiro. Present at Demo Day.

5
Finalist Development (23โ€“27 Nov)

Top 5 teams develop MVPs with Amazon Bedrock, mentored by AWS Solutions Architects.

6
Awards Ceremony (10 Dec)

Final pitches to judges. Winners announced! ๐Ÿ†

๐Ÿ“ฆ Submission Requirements

What to Submit

1. PartyRock App URLYour published, working PartyRock application
2. Project TitleClear, descriptive name for your solution
3. Problem StatementWhat Thai challenge does this solve? (200 words max)
4. Solution DescriptionHow does your AI app address the problem? (300 words max)
5. Target UsersWho benefits from this solution?
6. Demo Video2-minute video showing your app in action (optional but recommended)
7. Team InfoTeam members, roles, university/organization
Quality over quantity! A focused solution that solves one problem well will score higher than a broad solution that does many things poorly.

โš–๏ธ Judging Criteria

CriteriaWeightWhat Judges Evaluate
Problem Relevance25%Is this a real, significant Thai challenge? Clear problem definition?
AI Innovation25%Creative use of Gen AI? Meaningful AI integration (not superficial)?
Feasibility & Impact20%Can this be built? Would it actually help people if deployed?
User Experience15%Is it intuitive? Good design? Accessible to target users?
Presentation15%Clear explanation? Compelling demo? Team collaboration?

๐Ÿ† Prizes

๐Ÿฅ‡ 1st Place

เธฟ250,000

+ Incubation Opportunity + AWS Certification Vouchers + Amazon Echo/Kindle + Media Exposure

๐Ÿฅˆ 2nd Place

เธฟ150,000

+ AWS Certification Vouchers + Amazon Echo

๐Ÿฅ‰ 3rd Place

เธฟ100,000

+ AWS Certification Vouchers

๐Ÿ“„ Sample Capstone Structure

Example: "FloodGuard AI" โ€” Flood Early Warning System

ProblemThailand loses เธฟ50B+ annually to floods. Rural communities get warnings too late to evacuate safely.
SolutionAI-powered flood prediction using weather data + river sensors. Sends alerts in Thai via LINE/SMS 6-12 hours before flooding.
AI UsageClaude analyzes weather patterns + historical data โ†’ generates risk assessment + personalized evacuation routes.
Target UsersRural Thai communities in flood-prone provinces (Ayutthaya, Nakhon Sawan, Ubon Ratchathani).
ImpactEarlier warnings = more evacuation time = fewer casualties and property damage.
Use this structure as a template! Replace with your own problem, solution, and impact story.