multilingual-embeddings

Human-like memory for AI agents — semantic, episodic & procedural. Experience-driven procedures that learn from failures. Free API, Python & JS SDKs, LangChain, CrewAI & OpenClaw integrations.

Optimised BAAI/bge-m3 serving with dense + sparse + ColBERT embeddings, async dynamic batching and pipeline GPU inference

An implementation of knowledge distillation using contrastive learning (CLIP loss) to transfer knowledge from a large teacher model (OpenAI's CLIP) to a smaller multilingual student model (LaBSE) on a parallel English-Persian dataset.

Portuguese Food & Wine Guide — A RAG-powered conversational assistant for Portuguese cuisine, wine regions, and food-wine pairings. Built with Qdrant, Flask, PostgreSQL, Grafana, and GPT-4o as the final project for LLM Zoomcamp 2026.

English-Sinhala multilingual word embedding alignment resources

An end-to-end Nepali Budget Question-Answering system that combines LoRA fine-tuned multilingual LLM + Retrieval-Augmented Generation (RAG) to deliver accurate, context-aware responses about Nepal’s national budget.

基于 RAG 的本地论文问答系统：使用 ChromaDB + multilingual-e5 + DeepSeek，实现带引文溯源的中英双语 PDF 对话。

Lightweight HS code classifier using multilingual embeddings - POC for customs digitization

Self-hosted server for jinaai/jina-embeddings-v4 compatible with OpenAI's embeddings format.

A multilingual, topic-based language learning platform powered by generative AI. Learn real-life concepts (driving, digestion, emotion) across Spanish, French, Portuguese, and Italian, while sneakily introducing users to science, math, and well-being through natural everyday examples. Built for educational AI applications and gen-UX prototype

Indic-first embeddings you can actually ship. Documented sources. Weights: CC BY-SA 4.0.