yassir cheikh

• Design and industrialization of a GenAI agents platform integrated into OpenWebUI. • Development of GenAI pipelines in Python combining multimodal LLM/VLM and RAG architecture. • Advanced orchestration of agents with LangChain, LangGraph, and LlamaIndex; agent workflows and integration of business tools via MCP. • Secure deployment on local infrastructure with Docker and GPU inference. • Impact: automation of critical business processes and significant improvement in response quality and time. • Tech: Python, Faiss, Chroma, RAG, LLM, VLM, LangChain, LangGraph, LlamaIndex, MCP, Prompt Engineering, OpenWebUI, Ollama, Hugging Face, vLLM, Docker, GPU, FastAPI.

• Development of an intelligent chatbot based on RAG architecture to enhance educational resources. • Multi-program management with dedicated vector databases ensuring data isolation and response relevance. • Development of an automated document processing pipeline: analysis, structuring, vectorization, and indexing. • Integration of Azure AI Document Intelligence for document analysis and Azure AI Search for indexing. • Design of a complete and customized RAG architecture. • Deployment of the solution on the institution’s internal servers. • Impact: reduction in information search time and improved academic support. • Tech: Python, RAG, LLM, OpenWebUI, Azure AI Document Intelligence, Azure AI Search, Vector DB, Embeddings, NLP, LangChain, LLM.

Analyze and Enhance business processes. Creation of Backend APIs for the RUBIX process mining application.

Design and development of a real-time quality control platform for fuse boxes using a Deep Learning model and an Intel D435 camera for depth control (95% accuracy). Contribution to the improvement of metal combustion using Machine Learning and Clustering. Development of an intelligent legal agent based on LangGraph, orchestrating LLM calls and reasoning to analyze and answer labor law questions.

Conducted an in-depth analysis of banking transactions and developed a fraud detection model based on a graph neural network (GNN), increasing detection accuracy by 25% compared with conventional methods. Collected, analyzed and predicted weather data using PySpark, Hadoop and Kafka in real time. Design of a Grafana dashboard.