AWS Builder Center

Builder Center

AWS Builders Skill Sprint June 2026 – Building AI Agents with Strands SDK

AWS Builders Skill Sprint June 2026 – Building AI Agents with Strands SDK

Built five AI agent projects using the Strands Agents SDK as part of the AWS Builders Skill Sprint – April 2026. Since I didn't have access to Amazon Bedrock, I adapted the challenges to run locally using Ollama with TinyLlama while implementing conversational AI, custom tool calling, persistent memory, and a personalized AI assistant. This project helped me understand the core concepts of AI agents through hands-on development.

Prakash J

AWS Builders Skill Sprint – April 2026: Building AI Agents with Strands SDK

Introduction

As part of the AWS Builders Skill Sprint – April 2026, I explored the fundamentals of building AI agents using the Strands Agents SDK. The sprint introduced concepts such as conversational agents, tool calling, persistent memory, and intelligent assistants.
Since I did not have access to AWS Bedrock during development, I implemented the projects locally using Ollama with the TinyLlama model while preserving the same agent architecture and concepts.

Challenge 1 – First AI Agent

The first challenge focused on creating a basic conversational AI agent.

What I built

  • Configured a Strands Agent.
  • Connected it with TinyLlama through Ollama.
  • Successfully generated responses to user prompts.

Skills Learned

  • AI Agent Basics
  • Local LLM Integration
  • Prompt Execution

Challenge 2 – Tool Calling

In this challenge, I extended the AI agent by adding custom tools.

Tools Implemented

  • Calculator
  • Weather Tool
  • Age Calculator
The agent automatically selects the appropriate tool based on the user's request.

Skills Learned

  • Tool Calling
  • Custom Python Functions
  • Agent Decision Making

Challenge 3 – Persistent Memory

This challenge introduced the concept of memory.
Instead of cloud-based storage, I implemented a local JSON-based memory system that stores user information between conversations.
Example:
User:
"My name is Prakash."
Later:
User:
"What's my name?"
Agent:
"Your name is Prakash."

Skills Learned

  • Persistent Memory
  • JSON Storage
  • File Handling

Challenge 4 – Full AI Assistant

I combined all previous concepts into one intelligent assistant.
Features included:
  • Conversational AI
  • Calculator
  • Weather Information
  • Age Calculator
  • Persistent Memory
  • Streaming Responses
The assistant can answer questions while remembering previously stored user information.

Challenge 5 – Personal Memory Assistant

For the innovation challenge, I designed a Personal AI Memory Assistant.

Features

  • Answers general questions using TinyLlama.
  • Automatically stores user details.
  • Remembers previous conversations.
  • Retrieves stored information in future interactions.
  • Uses JSON for persistent local storage.
This demonstrates how AI assistants can provide personalized conversations without requiring cloud infrastructure.

Technologies Used

  • Python
  • Strands Agents SDK
  • Ollama
  • TinyLlama
  • JSON
  • Git
  • GitHub
  • VS Code

Challenges Faced

While the official challenges were designed around Amazon Bedrock, I adapted the implementation to work entirely on my local system using TinyLlama. This allowed me to understand the core concepts of AI agents, tool calling, and memory while experimenting in a local development environment.

Repository

GitHub Repository:
https://github.com/prakajayavel1994-code/nova-month-submission 

Conclusion

This skill sprint provided hands-on experience in building AI agents using the Strands SDK. I learned how conversational agents work, how to extend them with custom tools, and how to implement persistent memory for more personalized interactions. The experience also strengthened my understanding of local LLM integration and practical AI application development.
# aws-builders
Any opinions in this article are those of the individual author and may not reflect the opinions of AWS.
Enjoyed reading this content? Let the author know!

Your likes, comments, shares, and saves help creators reach more builders.

Loading recommendations

Loading article