At Mediology Software, I deliver backend systems, AI pipelines, and React admin panels for production products including Wishmaker, an ad-fraud platform, and SORTD Dynamic Config. For Wishmaker, I architected the backend for a multimodal pipeline that generates greeting cards, and added batch inference that cut model spend by ~50%.
I also implemented subscription flows across Google Play and the Apple App Store, with server-side webhooks managing production user entitlements. At IOTFY, I built a real-time monitoring platform for 3,000+ IoT devices and optimized MQTT service workflows, cutting execution time for critical operations by ~80%.
My projects include autonomous research agents that investigate flagged publisher domains for ad fraud, and an on-device article-detection app that processes 200+ PDFs daily. I’ve also built a RAG assistant that grounds a self-hosted LLM in an internal knowledge base and cites source chunks to reduce hallucinations.
I work across Python and TypeScript, with experience in Node.js, React, PostgreSQL, and production LLM systems. I’m focused on retrieval, structured outputs, multimodal pipelines, and keeping inference reliable and cost-conscious.

