BUDA Apps Portfolio

TradutorBUDA

Java-based Tibetan translation and text-processing application focused on Buddhist terminology workflows. It combines dictionary lookup, tokenization, transliteration, ranking/model services, and optional OCR-related utilities.

Tech: Java 21, Maven, SQLite, ONNX Runtime, tokenizers, Transformer (NllB, mBert), OCR utilities (CNN Model, NumPy, etc...)

• Tibetan term lookup and processing workflows
• Word-breaking and Tibetan text utilities
• Model/ranking and OCR-related optional integrations

GlossarioBUDA (Simplified)

Simplified glossary application retaining the core term management features without experimental OCR integration.

Tech: Java 21, SQLite, SWT, XML processing, SupaBase Real-Time Sync.

• Add, edit, delete, and search Buddhist terms
• Supports Tibetan, Sanskrit, Pali, English, Portuguese
• XML import, WhatsApp-friendly export, always-on-top quick view

Tibetan OCR — CNN & Transformer Research

Deep learning research project comparing lightweight CNN and CNN+Transformer hybrid architectures for Tibetan character OCR. Focused on replacing hand-crafted OCR pipelines with learned convolutional and attention-based feature extraction.

Tech: PyTorch, CNNs, Transformers, ONNX, INT8 Quantization, Gradio, Hugging Face Spaces, Mixed Precision Training

• 95.52% validation accuracy on 1,020 Tibetan character classes
• Lightweight CNN and CNN+Transformer hybrid comparison
• Interactive deployment using Hugging Face Spaces

Private Source Policy

The software is demonstrated publicly via videos and summaries while full source repositories remain private.

• Public: demos, features, architecture summaries
• Private: full implementation and proprietary assets
• Contact available for follow-up