Diana S.

I build NLP and LLM systems for text analysis, recommendation/ranking, LLM-generated output evaluation, and business decision support. With 15+ years of experience across Pharma, Healthcare, FinTech, and AI-driven product work, I help teams turn complex data, unstructured text, and model outputs into practical business decisions. My methodology for Evaluating Multi-Step LLM-Generated Content was featured on Towards Data Science, focusing on how to structurally evaluate goal-oriented AI outputs in complex customer journeys. My work combines NLP, machine learning, recommendation/ranking systems, personalization, LLM evaluation, and experimentation. I’ve built models and workflows for prediction, classification, recommendation systems, customer journey analysis, content targeting, engagement optimization, RAG, embeddings, topic modeling, semantic similarity, and A/B testing. I can help with: NLP and text classification LLM output evaluation frameworks RAG and semantic retrieval workflows Embeddings, clustering, and topic modeling Predictive modeling and classification Recommendation, ranking, and personalization systems A/B testing and experiment design Translating technical findings into business recommendations What sets me apart is that I don’t only build models — I evaluate whether they are useful, measurable, and connected to real business outcomes. Clear communication and structured thinking are important to me, so we can move efficiently from idea to working solution.