main projects needs:

PDF(docx, odt) to MD algorythm (e.g. docling, PyMuPDF4LLM, ...) local OCR LLMs (glm-ocr, ...) local oncurent small LLMs (GPT-OSS 20B, nemotron-cascade-2, granite4, ministral, ...) via Ollama, vLLM, llamacpp local dedicated translation LLMs (translategemma, ...) local embedding models (granite-embedding, USE, nomic-embed-text-v2-moe, ...) local classification models (Bert, ... ) local server for docker services (web+DB) and local LLMs