Data Scientist
Mountain View, California, Us
• Pioneered a suite of generative AI proof-of-concept applications using LLMs, including:o A Q&A system over Notion documents, using Retrieval Augmented Generation (RAG) with a Llama2-7b model, fine-tuned on 12,000 synthetic annotations.o A synthetic annotation generation bot using GPT 4 for scalable, instruction-tuned dataset generation, enhancing AI model fine-tuning.o A performance evaluation framework using system, RAG, and accuracy metrics.o A generative story AI for children's books using Stable Diffusion & Claude.o A SQL agent using LangChain for queries in supply chain analytics and specified a diagnostic agent counterpart that used Pandas dataframes and scikit-learn for machine learning.• Analyzed economic and governance risks for DeFi protocols (Curve, Lido, Maker, ERC-20), and created essential Lido protocol risk indicators which secured the startup's first customers.• Drove a UX redesign, significantly enhancing product appeal and user experience, directly contributing to initial customer acquisition and engagement.