At DealGround, I built an AI document-processing platform from the ground up, using Python services and LLM extraction to turn real estate documents into structured data. I raised extraction accuracy from 82% to 95% and reduced processing cost from 38¢ to under 8¢ per PDF.
I also built LLM-driven crawlers that use OpenAI models to navigate more than 300 real estate listing sites. They collect over 5,000 records per month at 0.001¢ per record, down from 50¢.
At GTPro, I led an eight-person development team and built the backend for storing and aggregating XRP Ledger donations. I introduced Agile practices and built CI/CD pipelines that moved releases from weekly to daily.
Earlier, I owned two HR-tech products end to end at VDX AI, where backend and database improvements cut p50 response time from about two seconds to about 0.3 seconds. As CTO at Quasar Systems, I led an 18-developer team delivering more than 80 client projects, and I continue to deliver custom software for clients as a self-employed Full-Stack Developer.
