software developer · AI systems · infrastructure · hinton, ab

I build practical systems out of messy real-world workflows

Software, AI tooling, self-hosted infrastructure, and cannabis retail data workflows. If it involves pulling order sheets, inventory dumps, and supplier catalogues into something that actually runs, I have probably already built it.

01 // projects

Things that actually run.

Production systems, internal tools, and side projects with real users, real data, or real infrastructure behind them.

01 / platform

Custom AI Chat

FastAPI + Next.js AI chat system with provider routing, streaming responses, prompt modes, workspaces, and test coverage. Built to own the stack end to end.

FastAPINext.jsStreamingTests
02 / tool

Header Hunter

Cannabis retail ordering workflow combining Greenline sales and inventory, AGLC catalogue data, supplier availability, and multi-store transfers into one consolidated order.

PythonPandasOpenPyXLGreenline
03 / web app

Plumbob Lineage

A Sims 4 family tree tracker built for a real user workflow. Import saves, map households and genealogy, and share trees with privacy controls. Free to use — paid tier for extra storage when the tree (and image count) gets big.

FlaskReactPostgreSQLRQ
04 / infra

MicroMountain Homelab

Multi-node Proxmox cluster running real services: Nextcloud, Authentik, Vaultwarden, n8n, Caddy, SearXNG, Jellyfin, Guacamole, and local LLM inference.

ProxmoxDockerCaddyLLMs
02 // systems

Stack, written down.

What I use to build and what I run it on. No marketing list — this is the actual tooling.

01
Backend
Python, FastAPI, SQLite / PostgresAPIs, data pipelines, and the glue between retail systems. Postgres for shared apps, SQLite for local tools, prototypes, and focused workflows.
Active
02
Frontend
Next.js, React, TypeScriptWhen the interface needs to be fast and the state needs to stay coherent across streams.
Active
03
Infrastructure
Docker, Proxmox, Linux, CaddyContainers on a 3-node cluster. Reverse proxy with automatic TLS. VMs for isolation, bare metal for performance.
Active
04
AI / LLMs
llama.cpp, local LLMs, OpenAI-compatible APIsSelf-hosted inference on Tesla P40s. Open WebUI, n8n hooks, and custom provider routing.
Active
05
Retail Data
Greenline, Excel, AGLC data workflowsPurchase orders, transfers, inventory dumps, and supplier catalogues turned into consolidated ordering logic.
Active
03 // work

What I do day to day.

Supervisor at Rocky Mountain Collective. The systems work is what makes the retail work possible.

supervisor · rocky mountain collective

Retail Ops & Data Systems

I supervise ordering for three stores under Rocky Mountain Collective. That means Greenline reports, Excel wrangling, supplier availability sheets, AGLC catalogues, and the occasional 200-line transfer order.

I built custom workflows around ordering, inventory reconciliation, and vendor data because the off-the-shelf tools don't understand how cannabis retail actually works. Less time clicking through portals, more time making sure the shelves are right.

GreenlineExcelAGLCPythonPandas