I build AI systems that handle real customers and real money. Every project in my portfolio is something I designed, built, deployed, and operate myself — no team, no agency, no templates. The phone number on this site rings an AI receptionist I wrote from scratch.
Ranked 1st–2nd in the entire Columbus district on overall performance. Learned to read people, close under pressure, and run operations with zero margin for error. This is where the sales instinct behind the AI agents comes from — I know what a real closer sounds like because I was one.
Transaction-driven work that sharpened negotiation and client management. The "sell for more / buy for less" frameworks I built into the AI negotiation systems started here.
Built the entire technical stack from zero: Shopify storefront, AI sales agent ("Rick"), voice AI receptionist, autonomous CEO operator, self-hosted PBX, email automation, Google Shopping feeds, Meta Ads campaigns, and a fleet monitoring dashboard. 83 products live, real paying customers, real revenue. Every system on this portfolio site runs here.
Built a multi-GPU LLM inference cluster (Tesla M40 + RTX 3060) on Proxmox. Custom CUDA build targeting Maxwell sm_52 hardware. Tuned a 27B parameter model to 19.97 tok/s by benchmarking layer placement across GPUs. This is production inference — it serves real traffic, not experiments.
PJ's Heating & Cooling (live client site with lead capture, SEO landing pages, SMS alerts). I Haul Junk Toledo (full local service site with city SEO pages). BizForge (one-shot AI business launcher). MarionMade (DoorDash-style food delivery platform with 2,003 AI-generated menu photos). Each project shipped end-to-end by me alone.
I don't hand off to another team. I own the backend, the frontend, the infrastructure, the deployment pipeline, and the monitoring. One person, one throat to choke, zero coordination overhead.
I don't build demos or POCs that need to be "productionized" later. Everything I ship is production-grade from day one — error handling, rate limiting, security headers, HMAC verification, KV persistence.
AI in front of paying customers needs guardrails. My sales agent has a hallucination guard that blocks fabricated prices before they reach a customer. My voice AI has persona boundaries. Trust is engineered, not assumed.
When data privacy, cost, or latency demands it, I run models on my own hardware. When cloud APIs are the right tool, I use them. I pick the architecture that fits the problem, not the one that looks impressive.
Every claim on this site is backed by a live URL you can hit. I don't say "deployed" until I've curled the endpoint and seen the response. The phone number on the homepage is a working AI receptionist — call it.
1–3 weeks from kickoff to production. Not because I cut corners, but because I've built the same class of system enough times that I know where the gotchas are before they bite.
I don't have a CS degree. I don't have former managers at FAANG companies who can vouch for me. I didn't learn AI in a classroom or a bootcamp.
What I have is a portfolio of live, running, inspectable systems that handle real customers and real money. Every project on this site has a URL you can visit, an API you can hit, or a phone number you can call. The work speaks for itself.
If you need someone with a Stanford pedigree and a long list of publications, I'm not your person. If you need someone who will go into the messy real problem and come back with a deployed system that works — let's talk.
Describe your project. I'll give you a straight answer on scope, timeline, and cost — no sales calls, no discovery workshops.
Get in touch →Or call (740) 223-5530 and talk to my AI receptionist. It's a demo of my work.