Director of Engineering for AI Engineering and Automation — and still writing the code. These are systems I've architected and built — multi-agent platforms at enterprise scale, autonomous AI products running as real businesses, and a self-hosted home AI command center with zero cloud inference. Not slide decks. Working software.
A multi-agent AI platform deployed across fulfillment sites nationally. Six production AI agents handle analytics over operational data, inventory intelligence, real-time operations control with 3D floor visualization, shipping, process mining with drift detection, and engineering decision-support. Built on an additive, "surgical insertion" architecture so it extends live systems without disrupting them.
A single secured portal consolidating ten internal operations applications behind enterprise identity and role-based access, with a modern front end and automated deployment. The connective tissue that gives one operation one coherent surface.
Large-scale system design: an event-sourced pharmacy ERP spanning ten domains and dozens of modules, plus an RFID + hardware-backed zero-trust authentication scheme for floor terminals. Shows the depth behind the production work.
A brand brief becomes a 30-second broadcast commercial in 10–20 minutes through a four-stage agentic pipeline: brand research, an AI creative-director concept pass, storyboard frame generation from product photos, and final video with native audio. Full subscription product with a token economy, brand profiles, auth, and 4K output.
A computational-geometry product that splits oversized 3D models into bed-fitting pieces with locking connectors and returns slicer-ready files. Geometry-aware seam finding, BSP partitioning, pegs/magnets/dovetails, and presets for common printers. A real algorithm with a real business behind it.
A full custom platform for a Phoenix integrated-wellness studio: a deployed dashboard app plus an auth backend, with live wearable-data integration pulling real biometric signals into a personalized client pathway — assessment, design, rhythm, evolve. A working AI platform delivered for a paying client.
NerdlyBuilds runs an automated content pipeline: cron-driven trend scraping, AI article generation, image generation, and a human review-and-approve step before anything goes live — all on Cloudflare's edge for a few dollars a month. The site is itself a working demo of automated AI content, plus a consulting front door.
A locally-run, multi-machine AI assistant that runs an entire command center with zero cloud inference. Two-tier local LLMs with speculative decoding, a full voice pipeline (wake-word → speech recognition → tool-calling orchestrator → streaming TTS) with multi-room audio and quiet hours, Home Assistant tool-calling, voice-driven 3D generation, a research-only paper-trading bot, and email rundowns that draft but never send. Triple-redundant disaster recovery with secret-guard tooling.
A self-hosted video generation and editing stack on DGX-class hardware: a text-to-video service exposed over an API and tool interface, plus an FFmpeg-based editor driven by LLM planning, with AI upscaling from 480p to 1080p. Cloud-grade capability, running entirely on local silicon.
Agentic automation, integration, or strategy — from someone who's actually shipped it to production. Let's talk about what your business needs.