Principal AI Engineer

I serve as the senior technical voice for AI systems, reviewing implementation, mentoring engineers on LLM design and evaluation, and partnering directly with clients and domain experts throughout delivery.

I built a production OCR and LLM invoice-extraction platform that processes thousands of utility invoices daily. I established its ground-truth evaluation methodology and drove context-engineering iterations that reached 97% aggregate field-level accuracy across several hundred fields.

I built secure MCP agent infrastructure, sandboxed subagents, semantic retrieval, telemetry, and hermetic reliability tests. I also delivered self-hosted conversational AI, proposal-generation, clinical voice simulation, embedded cold-chain monitoring, web services, and provider-outreach systems.

I delivered a custom Stable Diffusion LoRA end to end, including training-data curation, captioning, tuning across multiple runs, and release of the final adapter, which exceeded 3,500 downloads.

I built and containerized self-hosted products including a WebRTC screen-and-audio sharing application, a Matrix media bot using yt-dlp and ffmpeg, and a file storage and sharing platform with access controls and expiring links. I also completed an inherited PHP meetup platform with geolocation-based event discovery.

I was the sole engineer for a social photo-rating platform built with Flutter and Firebase. I developed a two-sided credit economy where users earned credits for ranking photo batches and used credits to distribute their own batches.

I designed the demographic targeting system for routing content by age, gender, and region, which also supported the paid advertising product. I owned the full stack, including the Flutter client, Firebase services, transaction ledger, profiles, and discovery feed.

Completed coursework toward a Bachelor of Science in Computer Science.