We resolved a major bottleneck in emerging market EdTech: scanned learning materials and legacy Sinhala font encodings (such as FMAbhaya and Kapitha) that standard text extractors cannot read.
Key implementations:
- Scanned PDF Detection: Automatically counts and slices scanned pages into 2-page batches, using Gemini Multimodal Visual OCR to transcribe text, math equations, and diagrams.
- Legacy Font to Unicode Converter: Integrated an automated translation prompt that reads visually encoded Sinhala characters and converts them into modern clean Unicode.
- Background Job Worker: Deployed startQueueWorker() to process file ingestion sequentially, preventing Gemini API 429 rate limits.
Log in or sign up for Devpost to join the conversation.