At Dynamic Safety LLC, I engineered Python ETL automation that reduced data-processing time by 67%, and created a tool that files payment and invoice documents using spreadsheet data. I also built a Power Automate workflow integrated with the Microsoft Graph API for collections outreach, with a human approval step.
At Prosper IT Consulting, I built a full-stack semantic AI chatbot with Next.js and TypeScript, integrating Voyage AI embeddings, vector search, and the Anthropic Claude streaming API in a RAG architecture. I also constructed its content ingestion pipeline, including Markdown chunking and rate-limited embedding generation.
For a client, I co-designed and co-coded a self-hosted tax automation pipeline using AllenAI OCR and a locally hosted Qwen model via LM Studio. It reduced processing time for a year's bank statements from eight hours to one; cloud deployment is underway.
I also built a Python and pandas pipeline analyzing CMS HCAHPS patient-experience data, with explicit handling of suppressed values, regression tests, and documented methodology. My work spans automation, AI integrations, financial systems, and data quality, backed by training in AI and software engineering.
