4 projects
rust-ai-serving-engine
Local GGUF model serving engine (pure-Rust Candle): chat-template generation, token streaming, session cache, and an OpenAI-compatible API with a Python binding.
rust-scholar-transformer
Real-time multi-source retrieval core engine (Rust) with a Python binding — arXiv/news/blog/YouTube/web fan-out, normalize, dedup, RRF rank fusion.
rust-ocr-transformer
Video-first Rust OCR pipeline — SSIM frame sampling + temporal dedup, emitting timestamped text/SRT for RAG ingestion.
rust-markdown-transformer
Deterministic, structure-preserving document → Markdown converter for vector DB / RAG ingestion (docx/pptx/xlsx/hwpx/pdf/html/markdown).
