Created on 2nd October 2024
•
Dakshata is a state-of-the-art AI-driven learning platform designed to provide personalized and dynamic educational experiences. Utilizing Google’s Gemini AI & real-time WebRTC technology, Dakshata adapts to each learner’s needs, adjusting the difficulty level based on real-time user responses and interactions. The platform delivers tailored learning content, ensuring that users progress at their own pace while continuously challenging them with scaled learning experiences.
While building the Dakshata platform, I encountered several significant hurdles, particularly related to integrating WebRTC and managing video recordings of students.
WebRTC Integration and Video Recording:
Integrating WebRTC for real-time video recording posed challenges. The initial setup was complicated due to the various configurations required for seamless video streaming and recording.
Solution: I spent considerable time reviewing documentation and troubleshooting issues with peer connections and media streams. After thorough testing and adjustments, I successfully implemented the video recording functionality.
Sending Video Content to Gemini:
Another major hurdle was sending the recorded video content to the Gemini API. Initially, the integration was not straightforward, especially with the need to communicate effectively between our application and Gemini's capabilities.
Solution: To resolve this, we developed a custom API for Gemini, which allowed us to send prompts and receive summaries of the video content more efficiently. This custom solution facilitated better communication and improved the overall functionality of the platform.
Handling Large Video Files:
We faced issues with sending large video files through the API, which led to timeouts and failures in the data transfer.
Solution: To address this, we implemented file chunking. By breaking down the video files into smaller chunks, we ensured that each part could be sent independently, thus avoiding size limitations and improving reliability.
These challenges ultimately contributed to enhancing the robustness of the Dakshata platform and provided valuable learning experiences in working with WebRTC and API integrations.
Tracks Applied (1)