Yossef Hassan

I build agentic AI systems that turn LLM outputs into reliable, validated workflows. My work spans multi-agent orchestration, RAG, tool calling, structured output validation, and production-minded persistence.

For Dinoora, I designed an end-to-end content pipeline of Story Planner, Script Writer, and Dialogue Writer agents. I built deterministic structural and semantic validators, provider-agnostic LLM integration, retry and idempotency handling, and provenance logging with Google Gemini as the production provider.

I also built a telecom customer-support chatbot using RAG, Qwen3-32B through Groq, ChromaDB, LangChain, and local sentence-transformer embeddings. It retrieves grounded context from FAQs, resolved tickets, and PDF guides through a Streamlit interface and CLI.

Alongside my projects, I teach applied AI and ML concepts as an AI Instructor. I enjoy translating technical material into practical, hands-on learning and building systems from architecture and prompt design through validation and persistence.