Mama AI Team

Mama AI is an intelligent maternal health assistant for African mothers. It offers local language voice support, health tips, reminders, and emergency aid to ensure no mother journeys alone.

Comment

Inspiration

We built Mama AI to make maternal health information accessible to pregnant and nursing women. Our goal was to create an AI-powered voice and text assistant that speaks both international and local languages and provides instant, trustworthy health guidance.

What it does

Mama AI lets users speak or type health-related questions in their preferred language. The system converts speech to text, translates it, retrieves relevant information using an AI model, and responds through text or voice. It helps users understand pregnancy care, nutrition, and maternal health practices in a simple, local-language format.

How we built it

We used a serverless, event-driven architecture powered by AWS:

  1. Frontend hosted on Amazon S3 and served via CloudFront.
  2. Users interact through voice or text on dev.auto-hive.site.
  3. AWS Lambda handles backend logic and connects all services.
  4. Amazon Transcribe converts audio to text.
  5. Amazon Translate translates text to the desired language.
  6. Amazon S3 stores translated text.
  7. Amazon Bedrock Agent Core generates intelligent responses.
  8. Amazon OpenSearch indexes and retrieves context with vector embeddings.
  9. Amazon Polly converts the AI response to speech for playback. ## Challenges we ran into
  10. Limited support for some local dialects
  11. Low accuracy of transcription in noisy environments
  12. Complex IAM role configuration for secure service communication
  13. Keeping latency low while maintaining scalability ## Accomplishments that we're proud of
  14. Built a working AI health assistant fully powered by AWS services
  15. Enabled local language interaction for improved accessibility
  16. Achieved end-to-end automation using a serverless design
  17. Integrated speech, translation, and conversational AI into one seamless experience ## What we learned We learned how to design and deploy an AI-driven multilingual system using AWS generative AI services. We gained hands-on experience with:
  18. Amazon Bedrock for intelligent conversation and response generation
  19. Amazon Transcribe, Translate, and Polly for speech and translation processing
  20. Amazon OpenSearch for semantic search and vector embeddings
  21. Serverless integration using AWS Lambda and S3 We also learned how to process and embed text data, manage IAM permissions, and deliver real-time voice interactions efficiently. ## What's next for Mama AI Team We plan to:
  22. Add support for more African languages
  23. Launch a mobile version for wider access
  24. Use Amazon SageMaker to train a local-language health model

Built With

  • bedrock
  • cloudfront
  • gnlp-api
  • lambda
  • node.js
  • opensearch
  • polly
  • react
  • s3
  • transcribe
  • translate
Share this project:

Updates