Software thatactually ships.
I design and build agentic systems and AI-native products — on seven years of shipping production software.
A model on its own is a demo.
What makes it software is everything around it — the routing, the tools it’s allowed to call, the data it’s allowed to see, and the layer that decides whether its answer is good enough to act on.
- 01Usera question, an instruction, a selection
- 02Router agentdecides what this actually is
- 03Sub-agentsauthor · verifier · domain specialists
- 04Tools · MCP · Datathe systems they're allowed to touch
- 05Guardrailsstructured output · eval · human-in-the-loop
AI & data
5RAG Knowledge Platform
Multi-modal ingestion, retrieval, and grounded chat
View →AI Writing Style Fine-tuning
DPO training pipeline for controlled generation
View →3D Model Generation
Text and image to 3D pipeline
View →AI Apparel Design & Quoting
Design generation plus tiered quoting
View →Unique Asset Generation
AI-generated per-order assets for Shopify
View →
Commerce
6Five published Shopify apps, plus a multi-tenant commerce platform running 15+ branded storefronts from one codebase.
Easy Edits
In-admin WYSIWYG editor for Shopify
View →Latch
Member-exclusive product locking & upsells
View →Pekel
GAAP-compliant subscription accounting
View →Drop Voice
Voice memo attachments for Shopify products
View →Taply
Branded NFC discount cards
↗Pocket Pitch
AI-generated pitch pages backed by NFC cards
↗
Free Custom LLM Agent Setup in 5 min (Claude + AGUI)
Want to build your own AI agent with a custom prompt and UI without spending hours configuring everything? In this video, I walk through how to set up a fully functional custom LLM agent in under 5 minutes using AGUI and Anthropic Claude. We'll clone an open-source GitHub project, configure a custom system prompt, customize the interface, and get everything running locally—for free. Whether you're building internal AI tools, coding assistants, customer support agents, or experimenting with autonomous workflows, this is one of the fastest ways to get started. In this video you'll learn: How to install AGUI locally How to connect an Anthropic Claude model How to customize the system prompt How to modify the UI for your own use case How to run everything locally in just a few minutes How to use the GitHub project as a starting point for your own AI applications This setup is perfect for developers who want a clean starting point for building custom AI agents without dealing with complicated infrastructure. If you're interested in AI engineering, agent frameworks, Claude, local development, MCP, or building production-ready AI applications, be sure to subscribe for more tutorials. Resources AGUI (https://github.com/ag-ui-protocol/ag-ui) Anthropic Claude API (Links in the video or pinned comment.) If this video helped you, consider leaving a like and subscribing—it really helps the channel and lets me keep creating free AI development tutorials. #AI #LLM #Claude #Anthropic #AIAgent #AGUI #OpenSource #GitHub #LocalAI #MachineLearning #ArtificialIntelligence #AIEngineering #AIDevelopment #PromptEngineering #Developer #Programming #Coding #SoftwareDevelopment #GenerativeAI #LLMApps #BuildInPublic #OpenAI #Python #JavaScript #TechTutorial
No More Complex CAD - 3D Modeling With AI & OpenSCAD
Honestly, I used to think 3D CAD modeling was a huge time sink, but then I started building with Claude and OpenSCAD. Turns out, having an AI draft your designs in SCAD language is pretty seamless, even for iterating. Create models and export them straight to STL files you can print.
Building AI-native software, agentic systems, and the infrastructure that makes them useful.
Shopify is one of the ecosystems I’ve spent the most time building in — five published apps and a multi-tenant commerce platform. It’s one ecosystem, not the whole picture.
I build AI-native software: agentic systems, the infrastructure they run on, and the products people actually use. That sits on seven years of production engineering — full-stack SaaS, distributed services, native mobile, embedded firmware — which is the part that makes agentic work ship instead of demo. Recently: a multi-tenant data and MCP platform, an agent deployment runtime, a browser-based CAD environment with an AI co-designer, and conversational analytics over live commerce data.
Agentic systems — routing, tools, skills, retrieval, and the permissions, policy, and tracing that make an agent safe behind a live endpoint.
Seven years of ordinary production engineering — SaaS, APIs, databases, payments, native mobile, firmware. The unglamorous half is why the AI half ships.
- Agent Orchestration
- Router + Sub-Agents
- Tool Calling
- MCP
- RAG
- Embeddings
- Fine-Tuning
- DPO
- Multimodal AI
- LangChain / deepagents
- Anthropic SDK
- Vector Databases
- Model APIs
- Retrieval Pipelines
- Evaluation & Guardrails
- Structured Outputs
- Streaming
- Human-in-the-Loop
- Training Pipelines
- TypeScript
- React
- Node.js / Express
- Python / FastAPI
- MongoDB
- SQL
- REST APIs
- WebAssembly
- Three.js
- Docker
- AWS
- Stripe
- ESP32
- STM32
- Arduino UNO Q
- MQTT
- NFC / NDEF
- Embedded C++
- iOS / Android
- Kiosk & Device Management
“Michael is one of the most irreplaceable developers I've had the pleasure of working with in my 6 years in eCommerce. Having scaled a Shopify agency from a team of 6 to 24 and revenues from $400k to $1.2m I know firsthand how HARD it is to find the right people. Michael is honest, skilled, and affable; he is a strong critical problem solver and a gifted technical developer.”
“Michael was excellent to work with. He outlined the goals of the project and hit every single one of them. Edits were fast and efficient. The work on the website looks great and I couldn't be happier.”
“Michael was extremely responsive, communicative and professional. His work was well commented and easy to understand. He worked hard to produce just what we wanted despite changing requirements.”
“Michael has been incredibly professional and helpful. We needed help with building a Shopify app and he stepped up to the task. He's also very communicative and responsive. Highly recommend!”
“Michael was super great to work with, if you are reading this right now wondering if you should hire him stop what you are doing and hire him... 11/10”