Любовь Авдеева

Python, Pydantic, LLMs, Ollama, prompt engineering, Streamlit, FastAPI. Focused on turning unreliable LLM outputs into production-ready systems.

Production LLM systems, structured output validation, multi-agent patterns, and better evaluation methods for generative models.

CreativeAudit — an LLM pipeline that evaluates ad creatives against brand guidelines with strict Pydantic validation and local inference.

Discussions about production LLM pipelines, Pydantic contracts, local inference, and reliable AI systems. Happy to chat about ad-tech + AI or review architecture ideas.