Craft Design Group · LLC

Hi, I’m Daniel. Engineering manager who codes.

I focus on agentic engineering — AI-native development workflows, sharing those practices across engineering teams, and shipping production AI features that hold up under real use. 14 years across architecture on AWS and Azure, frontend work in React, Vue, and Angular, and applications that run companies.

On the side I build open-source utilities for autonomous coding agents, custom operator software, and side projects when something interesting comes up. The work is below. Send me an email.

What I work on

Four overlapping lanes. Most projects touch more than one.

— 01

AI systems

Agent tooling, orchestration, and the infrastructure that lets autonomous things actually run.

— 02

Infrastructure

Control planes, observability, durable data, and pragmatic deployment choices.

— 03

Applications

Operator interfaces, internal tools, and full-stack products with real workflow depth.

— 04

Architecture

Technical direction, integration design, and the boring decisions that decide whether systems hold up.

Recent work

Things I’ve actually built and shipped.

Four projects. One client deployment, three open-source utilities. Click through for the build details.

01 Custom software

Studs Cycles POS

A shop-floor point-of-sale and repair-order system for an independent motorcycle shop.

One screen to triage every repair order in flight, a fast ticket builder, per-technician time tracking, and live profitability math — designed from the owner’s daily workflow up.

PocketBase + SvelteKitSingle-binary deployLive

Read the case study →

Studs Cycles POS — Workbench view
Plate 01 — Workbench view /work/studs-cycles

How I work

Build for use, not theater. Durable, cost-aware, fit-for-purpose.

  1. — 01

    Principle

    Match the constraints, not the trends.

    Cloud, self-hosted, hybrid, or local-first when it makes sense. The stack should fit the team, the risk, and the cost envelope — not whatever was on the front page of Hacker News.

  2. — 02

    Principle

    Build the surrounding machinery.

    A feature without runbooks, observability, and a way to roll back is half-shipped. The boring tooling around the code is what turns a working system into one that holds up.

  3. — 03

    Principle

    Utility should still feel good.

    Operator software is still software people use all day. Calm, clear, and sharp is a design value — not the same thing as flat or grey.

  4. — 04

    Principle

    Choose stacks that earn their keep.

    Small dependencies. Durable formats. Things that fit the shape of the actual problem. Add complexity when there’s a real reason; remove it the rest of the time.

Contact

If something here caught your eye, send me an email.

Agentic engineering, AI-native solutions, and sharing those practices with engineering teams. Architecture and implementation for software that has to actually run, not just demo. Most of my work is hands-on.