I enjoy building things — mostly with my hands at a workbench, but more recently with my fingers at a keyboard.

By day I work at Qualcomm (via the Edge Impulse acquisition), helping enterprise customers bring machine learning to edge devices. By night I'm usually in the shop, running the CNC, or poking at something that probably shouldn't be automated but is going to be anyway.

LLM Things

• Modelo — Native macOS client for running inference against local and cloud LLMs. Connects to LM Studio over Tailscale, OpenRouter, and any OpenAI-compatible API. Streaming chat with token metrics, agentic tool use (Firecrawl + MCP), model load/unload from the picker, and a menu-bar mini chat. SwiftUI + SwiftData + Swift Charts, zero third-party packages. Named after the beer 🍋‍🟩 the logo is a lime.

• Strafe — Native macOS/iOS chat client for LM Studio. Streams responses in real time via SSE, supports vision models, and persists conversation history with SwiftData + CloudKit.

IOS

• TacoScout — iOS app that finds nearby taco spots using SwiftUI, MapKit, and Google Places. I love specialized tools, and that principle applies here, a simple application that does one thing... FINDS TACOS

• ScatTrax — iOS app using CoreML and iNaturalist data to identify animal species from scat. Vermont has a lot of wildlife and I have questions.

Web

• hideout — Scans real Northeast terrain data to rank hollows, dense canopy, and defensible hidey-holes. Science in service of antisocial tendencies.

Extra-Curriculars

• Peregrine Design Werx — My hobby slash side business doing custom CNC woodworking, laser engraving, and 3D printing. Shapeoko 5 Pro, xTool P2 CO2 laser, Prusa Core One and a pile of sawdust.

Learn by doing. Break shit. Experiment. Use the best tool for the job — whether that's a chainsaw, a router bit, or an LLM.

The woods, Cambridge, VT