AI ENGINEER & SERIAL FOUNDER
PORTLAND, OR // AVAILABLE NOW
$10+ years shipping production ML, voice & video AI, computer vision and go-to-market infrastructure — from healthcare AI to agentic workflows.
$./talk_to_my_ai_assistant --live
Describe your challenge on a live video call with my AI assistant. You'll need an invite code to get in.
No invite code? Book a call instead →
Not slideware. Production systems with real users, real latency budgets and real bills.
Production AI for clinical documentation, NLP-driven compliance and automated billing. Built a custom MCP server wiring Guident into Stedi for insurance eligibility and claims.
AI-powered 1:1 video matchmaking that replaces cold outreach with high-intent meetings. 500+ matches shipped at a 94% satisfaction rate.
Draw a fire perimeter over Portland and it runs the whole spatial workflow: PostGIS ST_Buffer cuts GO/SET/READY zones as true annuli on Oregon North, a spatial join rolls up real 2020 Census population per ring, GiST KNN finds the nearest shelter that still has capacity, and any route crossing the fire is rejected and re-solved. 592 block groups, 815,428 residents, 61 real NIFC perimeters. ArcGIS Maps SDK + FastAPI/Shapely + ASP.NET Core over PostGIS on Neon.
Gives Claude real hands in your own browser. A Chrome extension + MCP server that sends human-shaped input — curved mouse paths, real drags, isTrusted keystrokes — through Chrome's pipeline, with safety checkpoints before anything irreversible. 35 tools, MIT, no telemetry.
Automated outbound calling with AI voice agents. Conversational agents dial, hold a real conversation and follow up on their own — turning a manual call list into an automated pipeline.
“Tell us the plan. We'll find the ride.” An SMS + voice brokerage that matches party bus, limo and coach operators to your event. Text the details, get vetted locked-in quotes in ~10 minutes. No app, no surprise fees, card details never shared with operators.
A decentralised protocol for autonomous agent compute — any AI agent can act, pay and be held accountable with no human in the loop. Four primitives: signed agent identity, job routing, EVM escrow settlement in FSO, and permanent on-chain receipts. Node client in Rust + Tauri, VM-isolated with a scoped credential vault.
Real-time voice pipeline stitching streaming ASR, LLM inference and TTS into one low-latency architecture.
Real-time lip synchronisation and expression enhancement running on dual-GPU acceleration on GCP.
Fully local voice assistant with real-time streaming and page-level RAG. Faster-Whisper, Kokoro TTS, ChromaDB and sentence-transformers — zero cloud dependencies.
Anti-cheating system for online exams — real-time gaze tracking and anomaly detection from a plain webcam feed.
Sign language translator built on 1D-CNN architectures for gesture recognition over a custom dataset.
Multi-agent video generation pipeline — scriptwriter, creative director, image director, animator and editor agents on FastAPI + Gemini, rendering finished content end to end.
Prophet forecasting models, product-intelligence scrapers, email verification tooling at scale, voice-activated Chrome extensions and more.
RAG pipelines, agentic workflows and MCP integrations across Claude, GPT, Bedrock and Llama.
Gaze tracking, gesture recognition, lip sync and real-time video analysis running in production.
End-to-end model development in PyTorch and TensorFlow. Data pipelines through to deployment.
Low-latency voice pipelines, real-time avatars and streaming ASR + TTS + LLM inference.
Clinical documentation, automated billing and insurance eligibility automation at scale.
React, Node.js, TypeScript, Django, Flutter. Prototype to shipped product with CI/CD behind it.
AWS, GCP, Azure, Docker, Kubernetes, Firebase, Supabase. Deployments that survive traffic.
Go-to-market stacks with Clay, Hunter.io and GoHighLevel. CRM automation and AI outreach.
Time-series forecasting with Prophet, product intelligence and data-driven decision systems.
Not a wish list — every one of these is load-bearing in something that shipped.
Start with my AI assistant on a 5-minute video call, or skip the queue and book time with me directly.