At ScottishPower, I build and deploy GenAI/RAG platforms using OpenAI LLMs and FAISS across thousands of operational documents, reducing retrieval time to about five seconds. I also developed a conversational AI copilot that improved agent KPIs by 25% and analytics pipelines that lifted customer experience scores by 15%.
I've built LangGraph agent workflows, optimized prompts with DSPy and LLMLingua-2 to reduce LLM infrastructure costs by 30%, and delivered forecasting and risk models that reduced escalations by 20%. I take solutions into production with Python, FastAPI, Docker, Kubernetes, CI/CD, Azure, PostgreSQL, and Alembic.

