At Whitestone Associates, I built a multimodal RAG pipeline that reads historical maps, aerial imagery, city directory records, and environmental database results to flag likely recognized environmental conditions for engineers. I also split retrieval into distributed components so a failed step wouldn't stall the rest.
I defined environments in Terraform, deployed them on Kubernetes, and ran the pipeline and report assistant on AWS with updates shipped through CI/CD. Engineers use the Python report assistant to review flagged sites, and I reworked its screens based on their feedback. I own the firm's generative AI work, selecting and building production tools that save engineers time.
Before Whitestone, I started Stealth Startup, where I built a Python document assistant prototype for small healthcare and professional-services firms. It used RAG and a vector database to return cited answers, with an AI agent that drafted summaries and follow-up emails. At Keystone Strategy, I delivered data science and machine learning consulting, partnered on computer vision and NLP models for information retrieval, and advised clients on AI data privacy.
