This is a screening assignment for our developer position. The task involves building a web application that enables voice-based interaction with SEC filing summary data using RAG (Retrieval Augmented Generation) and speech-to-speech conversion.
Financial professionals often need quick, hands-free access to SEC filing information while multitasking or on the move. This assignment focuses on building a practical application that combines speech RAG and speech-to-speech technologies to provide natural voice-based interactions with SEC filing data.
Build a web application using Reflex.dev that allows users to query SEC filing information using voice input and receive spoken response. The application should utilize RAG to retrieve relevant information from the provided database of SEC filing summaries and generate concise, contextually relevant responses.
The app should work similar to https://www.turtlee.in/voice
The dataset consists of summaries of SEC filings along with some metadata. RAG must be implemented on top of summaries.
Link to dataset: Click Here
- Implement speech-to-speech conversion for user queries
- Create a RAG system using the provided SEC filing summaries database
- Generate natural language responses using an LLM
- Maintain conversation history and context
- Provide visual feedback during voice interaction
- Use Reflex.dev as the primary framework
- Build speech-to-speech processing engine
- Create RAG pipeline from the provided summary database
- Integrate with an LLM API for response generation
- Include a fallback text interface
- App should work for 1 user flawlessly
- Write clean, well-documented code
- Include comprehensive type hints
- Follow Python PEP 8 style guidelines
- Implement proper logging and monitoring
- Document RAG implementation details
- Code quality and organization
- Speech processing accuracy
- RAG system effectiveness
- Error handling and edge cases
- Database query optimization
- Speech recognition accuracy
- Voice response naturalness
- Conversation flow
- Handling of ambient noise and accents
- Fallback mechanisms
- Accuracy of retrieved information
- Relevance of responses
- Response conciseness
- Context maintenance
- Citation accuracy
- Response time optimization
- Resource usage efficiency
- Query optimization
- API usage optimization
Your submission should be a ZIP file containing:
- Complete source code of the application
- README.md with:
- Setup instructions
- API configuration steps
- Voice interaction examples
- Database connection guide
- Performance considerations
- Requirements.txt or similar dependency file
- Testing scripts and documentation
Send your submission as a compressed ZIP file to info@alphanome.ai
- Implement appropriate rate limiting for all APIs
- Include error handling for speech recognition failures
- Document any assumptions about the summary data format
- Consider implementing a caching mechanism
- The assignment should take approximately 5-10 days to complete
- Focus on core functionality first, then add improvements if time permits
If you have any questions about the assignment, please email info@alphanome.ai