My Partyrock journey!
A learning opportunity..
Published Mar 7, 2024
Inspired by Master Gardeners:
My journey began when I first learned about Amazon Bedrock while it was still in preview. I was eager to test its capabilities, but the limited free tier restricted my initial exploration. However, upon hearing about PartyRock and the hackathon, I knew I had to participate.
Project Impact:
My project is inspired by the dedication and passion of Master Gardeners who contribute to the community through sustainable gardening. This app aims to empower gardeners by leveraging AI for improved planning and yield.
Key Features:
- Climate-Specific Plant Recommendations: The app accurately suggests flowers and vegetables best suited to gardener's local weather conditions, ensuring compatibility and success.
- Planting and Harvesting Schedules: We provide precise timing for propagation and seedling planting, crucial for maximizing the yield.
- Optimized Plant Combinations: The app recommends ideal plant combinations, further enhancing a gardener's success potential.
- Eco-Friendly Pest Control: Application offers environmentally friendly methods to combat pests, protecting your soil and contributing to a healthy ecosystem.
- Centralized Information Source: Eliminates the search struggle! This app offers a comprehensive solution, saving gardeners valuable time and effort.
Alternative Development Approach:
While participating in the PartyRock hackathon sought a competitive edge, the real value for me was the extensive knowledge gained. I had the opportunity to experiment with various foundational models, all free of charge. If I were to actually this application, here is how my dev stack would look like.
- Front-End: Single-page application, CSS, Angular
- Back-End: AWS Lambda
- Lambda - Model integration: Boto3 library for invoking Bedrock models, potentially a knowledge base and an agent.
- Content Delivery Network (CDN): AWS CloudFront
- Hosting: AWS Amplify/S3 backed.
- Authentication: AWS Cognito
- Domain Name System (DNS): AWS Route 53
- Amazon Polly - For building a YouTube video
To begin, I started cataloging available models in PartyRock. I systematically evaluated each option for specific tasks, optimizing results by adjusting top P, temperature, and configuration values. This process also significantly enhanced my understanding of prompt engineering. In the absence of PartyRock, I would have utilized existing Master Gardener content, ensuring proper attribution of resources, in the form of a Response Augmentation Generation (RAG) model via Amazon Bedrock's Knowledge base feature.
You can checkout the video here - https://www.youtube.com/watch?v=4INtYmcsJ9o
Conclusion: Partyrock hackathon was a great learning experience which proves that Amazon/AWS is dedicated to investing in Generative AI and making world a better place for everyone.