Skip to content
View successbrian's full-sized avatar

Block or report successbrian

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
successbrian/README.md

Brian Lathe

Applied AI systems builder — Minneapolis area, Minnesota

I build and operate autonomous AI systems that run around the clock on hardware I own. My main project, SuccessBrian OS, is a working multi-agent operating system for a one-person business: it watches a fleet of machines, processes its own work queues, prepares decisions for human ruling, and keeps a permanent, queryable record of everything it does — 24 hours a day, whether I'm at the keyboard or not.

See it running

  • Live operations dashboard: https://successbrianvps.com/pm/ — the system's own public window: work queues, decisions, and system health, updated by the machines themselves.
  • Website: aiprofits2025.wixsite.com/successbrian-os — the SuccessBrian OS site.
  • Blog: successbrianhub.substack.com — I write about building and operating this ecosystem: what works, what breaks, and what it teaches.
  • Flagship repo: successbrian-os — 300+ commits of deterministic Python, PostgreSQL, and local LLM inference. No framework-of-the-week; the boring, reliable parts are code, and small local models are used only at defined judgment points, with their outputs validated before anything acts on them.

What the system does

  • Runs a real fleet: 10+ machines — mini PCs, X79 and DL380 servers, and a cloud VPS — running local inference (35B-class models) on my own hardware.
  • Keeps its own memory: a knowledge graph of 2,500+ nodes plus a decision stream that records what was decided, by whom, and why — so nothing depends on anyone's recollection.
  • Repairs itself first: watchers restart failed workers, silent processes get hunted down by their own logs, and only what the system genuinely can't fix reaches me.
  • Knows who decides: the software proposes, ranks, and prepares. I rule. That separation is the design, not a disclaimer.

How I work

Deterministic code first; models where judgment is actually needed; every claim the system makes carries its evidence. I document decisions the way I want them read — for intent, so the next reader (human or machine) understands what was meant, not just what was said.

Currently

Open to full-time roles in applied AI, AI operations, and agentic systems — especially teams putting local or hybrid LLM systems into real production work. The best way to evaluate me is above: the dashboard is live, and the code is public.

📍 Minnesota · US Navy veteran

Popular repositories Loading

  1. prime prime Public

    HTML

  2. project_corona_tracker project_corona_tracker Public

    Forked from adrianhajdin/project_corona_tracker

    This is a code repository for the corresponding YouTube video. In this tutorial we are going to build and deploy a corona tracker application. Covered topics: React.js, Chart.js, Material UI and mu…

    JavaScript

  3. nucampsite nucampsite Public

    JavaScript

  4. mlm-refferal-tracking mlm-refferal-tracking Public

    Forked from rg-rupeee/mlm-refferal-tracking

    Multilevel Marketing Refferal Tracking system and assigning rewards based on 3 layers deep referral

    JavaScript

  5. freqtrade freqtrade Public

    Forked from freqtrade/freqtrade

    Free, open source crypto trading bot

    Python

  6. crypto-js crypto-js Public

    Forked from brix/crypto-js

    JavaScript library of crypto standards.

    JavaScript