Relationships
Hardware, software, user flow, integrations, ownership, and operations, tied together before the first line of code.
Founder profile
I help smaller businesses with software, automation, and AI that works in real operations.
My strength is connecting idea, system, and practical implementation: from an unclear task to a first version that can be tested and used right away.
Read the full technical profile
Thinking model
System before feature
Process, interface, data, machine, and failure modes. Map first, then build.
Version fast
Small builds, real feedback, fewer assumptions, and quick restructuring along the way.
AI with ownership
Codex, Claude, ChatGPT, prompting, and review loops, with human direction the whole way.
Operable structure
Modularity, orchestration, documentation, maintenance, and room for the next iteration.
Position
Down from the cloud, onto the desktop.
My stance on AI isn't that it should be avoided. It's about where it lives, who owns it, and what it costs to run.
Most of the industry builds AI as a call to somebody else's server. That's fast to get started with, but it makes the product dependent on an API key, a price that can change, and a logging policy you didn't write yourself. I build it the other way around: the model should run locally wherever possible, on the machine that actually needs it, and the cloud becomes the exception instead of the default.
That's not a technical preference for its own sake. It's three things I take seriously whenever I build something for someone: who owns the result, where the data actually travels, and how much compute gets spent on a task that never needed it.
The model lives on the machine
Ollama, local inference, and a clear line for when a task genuinely needs a cloud model, and when it doesn't.
The tool should outlive me
Code, data, and decisions that can be inspected, exported, and changed, without depending on a subscription staying active.
Data only travels when it has to
The simplest privacy guarantee is still that data never leaves the machine. That's the starting point, not a feature you switch on.
Match the model to the task
A small local model for a small task isn't just cheaper. It's also a rejection of routing everything through a huge cloud model that was never proportional to the job.
Technical range
Bridge between digital workflows and physical systems
Stack isn't my identity. It's the tool that gets a system running.
PLC, HMI, robotics cells
TIA Portal, TwinCAT 3, OPC UA, sequence logic, PID, sensors, and fault handling.
Edge interfaces and prototypes
ESP32, ESP32-S3, Raspberry Pi Pico, MicroPython, GPIO, displays, and motor drivers.
Web apps, internal tools, product flows
Frontend, static-first delivery, Node.js tooling, content systems, and deployment.
Agentic workflows and technical acceleration
Codex, ChatGPT, Claude, Cursor, n8n, context management, and review loops.
Good fit
When the gap is between idea, system, and operations
First versions
Scope, system sketch, prototype, release, and learning along the way, in that order.
Messy workflows
Data routing, n8n, handoffs between people, automation, fewer manual steps.
Technical product shape
Information architecture, systems UI, credible frontend, operational clarity.
Current
Focus now
- DeskDude ServicesWebsites, AI workflows, automation
- Restaurant TrainingSeparate product, app flow, training
- Technical notesAI, workflows, systems building
Best for
Typical work
- ClarifyingRaw ideas, scope, buildable steps
- PrototypingFirst version, feedback, iteration
- BridgingMachine context, interface, data flow
Contact
Have a scoped build?
Send a short brief. Context matters more than polished writing.
Start dialogDeep dive
Full technical profile
Overview for work where software, automation, AI, and physical systems knowledge meet.
Automation
PLC, Ladder, Function Blocks, TIA Portal, TwinCAT 3, OPC UA, SQL, IO, sensors, actuators, buffer logic, RFID, fault handling.
Robotics and regulation
DENSO, WinCaps III, PACScript, palletizing, robot/PLC coordination, PID, Ziegler-Nichols, signal analysis, deterministic control.
Embedded
ESP32, ESP32-S3, TTGO T-Display, Raspberry Pi Pico, MicroPython, GPIO, TFT, motor drivers, sensors, edge-device logic.
AI and software
Codex, ChatGPT, Claude, Cursor, Ollama, OpenAI APIs, n8n, React, Node.js, prompting, context, review, deployment.
Product UI
Information architecture, compact navigation, technical copy, high signal, and low visual noise.
Working pattern
Understand the system, build the version, test the constraints, remove friction, document the next iteration.