Independent research node onlineDetroit area / open workbench / 2026

Chuck Gamburd / Applied AI Systems Designer

Open-source AI research,built into systems.

I turn messy, real-world problems into working software, research protocols, agent infrastructure, and tools with visible evidence. The work spans AI, vehicle interfaces, field service, markets, generative geometry, and the systems connecting them.

Research signal matrix

BRN::FIELD_TO_MODEL

01 Agent systems92.0
02 Applied AI86.0
03 Evidence96.0
04 Interfaces76.0
05 Field systems82.0

36

records

07

systems

OPEN

method

Current instruments

The work is the proof.

All projects

Android systems / vehicle workbench

Silverado Cortex

A phone-first cyberdeck environment for a 2016 Silverado: one canonical state graph, modular adapters, local knowledge, Neuron retrieval, an app bridge, and deliberately gated vehicle-facing integrations.

Active · v0.5.0 live-test artifact

The Web Deck debug build follows a verified v0.4 App Bridge foundation with scoped discovery, SQLite v4 persistence, offline tests, emulator proof, and no broad package or network permissions.

KotlinAndroidSQLiteLocal-firstSafety gates

Generative geometry / MCP

TurtlePen

An integer-exact lattice for AI-authored diagrams: measurement before placement, whole-quadrant geometry, and severity-ranked collision findings instead of silent resizing.

Open source · active prototype

424 tests, 39 MCP tools, zero runtime dependencies, and a pure core separated from the local stdio transport and viewer.

JavaScriptMCPGeometryZero dependencies
Open repository

Trading research workstation

BrainnStation

A desktop trading desk that brings market discovery, stock detail, journaling, AI-assisted analysis, capture, OCR, speech, visual reading, and backtesting into one operating surface.

Active · v2.0.9

Electron 39, React 19, TypeScript, a Python sidecar, and SQLite power a dense workstation built around decisions rather than disconnected widgets.

ElectronReactPythonSQLiteMarket data

Research field map

One practice. Multiple realities.

Research is organized by problem domain, but the method stays consistent: identify the real constraint, make the boundary explicit, build a testable structure, and keep the evidence.

Open all research

01

AI systems

Semantic compilers, agent operations, context recovery, orchestration.

02

Interfaces

Generative geometry, spatial UI, benchmarks, and instrument surfaces.

03

Applied systems

Cortex, HVAC field systems, vehicle research, and real constraints.

04

Markets

Trading workstations, strategy, journals, and evidence packs.

05

Creative systems

Stream direction, creator education, games, and characters.

06

Security & evidence

Governance, scanning, bounded iteration, and claim discipline.

Operating principles

Research that survives contact with reality.

Evidence before confidence

Claims stay proportional to what was actually inspected, tested, or observed.

Field constraints are design input

The truck, the rooftop unit, the trading desk, and the operator all change what good software means.

Truthful failure states

A tool should say what happened, what did not happen, and what still needs proof.

Source is part of the conversation

Public code, experiments, forks, and working notes show the path—not just the polished result.

View GitHub