Available

Systems, not just screens.

A case-study layer for product systems, operational software, AI-connected workflows, and open-source engineering infrastructure. Each project is framed by the problem, architecture, decisions, proof, and outcomes because senior engineering should leave a trail.

Case files

Architecture, decisions, proof.

JPope Tax Portal
012024Live

JPope Tax Portal

Enterprise client portal for tax consultancy

Impact lineA portal that behaves like operational software, not a veneer over back-office chaos.
SurfacePortal + docs + comms
ModeRealtime-ready
FocusSecurity + clarity
System topologyInput -> core -> outcome
01
Input surfaceClient + staff intake

Documents, messages, requests, and tax workflow state enter through one portal surface.

02
System coreRole-aware workspace

Realtime data sync, access boundaries, document status, and communication records stay connected.

03
Output layerTrust-heavy client UX

Clients and staff see a coherent operational layer instead of fragmented back-office tooling.

Control signals
RBAC boundariesDocument-state transitionsRealtime sync health
Proof telemetry
Primary riskAccess clarity
Data modeRealtime
OutcomeWorkflow trust
Evidence mediaScreenshot + artifact telemetry
JPope Tax Portal screenshot
A full application surface with client workflow, trust posture, and operational controls visible in one frame.
Artifact typeClient portal UI

Shows productized tax consultancy operations rather than a static service brochure.

Open motion artifact ->
Role-aware UXDocument workflowClient trust surface
Artifact telemetryInspection trace
Capture classPortal workspace
Proof posturePrivate product UI
System signalRBAC + docs
Problem

The business needed one secure client surface instead of fragmented communication, ad hoc document exchange, and manual operational follow-up.

Architecture map
01Role-based access
02Realtime data sync
03Document workflows
Build decisions
  1. Built a role-aware client portal with unified communication and document workflows.
  2. Used a realtime-capable frontend architecture so the product felt operational, not brochure-like.
  3. Focused on making enterprise complexity legible to both staff and clients.
Proof signals
  1. Enterprise-flavored interface discipline, not just marketing-site polish.
  2. Shows ability to design around access control and business workflow complexity.
  3. Demonstrates product thinking in a regulated, trust-heavy environment.
Outcomes
  1. Unified client communication surface
  2. Reduced manual back-office overhead
  3. Production-ready enterprise UX
JPOPE Tax Consultancy
022026Live

JPOPE Tax Consultancy

JamiePope.com commercial real estate tax website

Impact lineA specialist tax consultancy turned into a public planning interface owners and advisors can route through.
Experience30+ yrs
Service areas23
Planning lanes6
System topologyInput -> core -> outcome
01
Input surfaceOwner + advisor questions

Acquisition, ownership, depreciation, credits, appeals, disposition, and CPA handoff questions enter through focused service and resource routes.

02
System coreCRE tax routing model

The site maps pressure points into planning lanes, service clusters, resource assets, newsroom notes, and consultation paths.

03
Output layerCPA-ready inquiry path

Visitors can understand the likely planning lane, review supporting material, and move toward a focused advisory conversation.

Control signals
Service taxonomyPlanning-lane routingResource freshnessConsultation CTA continuity
Proof telemetry
Experience30+ yrs
Services23
OutputCPA-ready
Evidence mediaScreenshot + artifact telemetry
JPOPE Tax Consultancy screenshot
A production consultancy site organized around CRE tax lifecycle planning, planning lanes, service routes, and advisor-ready next steps.
Artifact typeBusiness website system

Shows information architecture, positioning, content modeling, and conversion paths for a specialized tax advisory practice.

Service taxonomyPlanning-lane modelCPA-ready conversion path
Artifact telemetryInspection trace
Capture classPublic website
Proof postureLive client site
System signalCRE tax IA
Problem

JPOPE needed a public web presence that could explain specialized CRE tax strategy without flattening it into generic service copy or forcing owners to self-diagnose complex tax questions.

Architecture map
01Service-route taxonomy
02Lifecycle planning framework
03Resource + newsroom content model
Build decisions
  1. Structured the site around decision pressure, planning lanes, service clusters, resources, and advisor-ready outputs.
  2. Turned niche tax expertise into a scannable public interface with direct routes into services, resources, newsroom notes, and consultation flow.
  3. Used the content architecture to connect commercial positioning with technical credibility and conversion clarity.
Proof signals
  1. Live production website with a full service taxonomy and advisor-friendly planning model.
  2. Frames 30+ years of CRE tax experience across 23 service areas and six planning lanes.
  3. Demonstrates ability to build a content-rich business site as a structured decision system, not a static brochure.
Outcomes
  1. Makes specialized CRE tax work easier to navigate
  2. Connects service strategy with conversion flow
  3. Turns domain expertise into a structured public system
TaxScout.app
032025Live

TaxScout.app

AI-powered commercial real estate tax optimization

Impact lineTurns specialized tax engineering into software people can actually use.
DomainCRE tax workflows
AI layerMCP-connected
Product moveDecision support
System topologyInput -> core -> outcome
01
Input surfaceProperty + tax inputs

CRE assumptions, depreciation context, exchange scenarios, and user goals enter the model.

02
System coreDomain calculation engine

Tax workflows become structured decision paths with AI/MCP access for agent-assisted analysis.

03
Output layerInvestor decision support

Users get understandable optimization paths instead of raw tax logic and disconnected spreadsheets.

Control signals
Scenario boundariesMCP access policyCalculation explainability
Proof telemetry
DomainCRE tax
AI pathMCP
UX modeGuided analysis
Evidence mediaScreenshot + artifact telemetry
TaxScout.app screenshot
Commercial real-estate tax logic presented as an inspectable decision-support surface.
Artifact typeCRE tax workflow

Pairs domain-heavy calculations with an AI/MCP-accessible product model.

Open motion artifact ->
Domain modelingGuided analysisAI access layer
Artifact telemetryInspection trace
Capture classDecision surface
Proof postureLive domain app
System signalAI/MCP model
Problem

A domain-heavy tax workflow had to become understandable, actionable, and productized for users who care about financial outcomes more than internal complexity.

Architecture map
01AI agent publishing surface
02Tax calculation workflows
03Multifamily domain modeling
Build decisions
  1. Translated niche tax strategy into guided product workflows.
  2. Connected domain logic to AI/MCP access so the platform could participate in agent-driven workflows.
  3. Designed the frontend as a decision-support surface rather than a generic dashboard.
Proof signals
  1. Shows you can productize high-complexity domain knowledge.
  2. Signals AI integration as part of system design, not as a decorative feature.
  3. Combines frontend execution with business-model awareness.
Outcomes
  1. Turns niche tax logic into a usable product
  2. Demonstrates AI + domain depth
  3. Shows product thinking beyond pure frontend
MAINFRAME
042025Live

MAINFRAME

Open-source bash stdlib for AI agents

Impact lineInfrastructure thinking, expressed in the most unforgiving runtime: pure shell.
Functions4,000+
RuntimePure bash
Use caseAI agent systems
System topologyInput -> core -> outcome
01
Input surfaceShell + agent calls

Scripts, prompts, files, APIs, and terminal state flow into a bash-native standard library.

02
System corePure bash primitives

JSON, HTTP, CSV, memory, RAG, embeddings, and orchestration live without heavyweight runtime assumptions.

03
Output layerComposable agent ops

AI-native shell workflows gain reusable infrastructure instead of fragile one-off glue.

Control signals
Runtime portabilityFunction-level compositionAgent memory boundaries
Proof telemetry
Functions4,000+
RuntimeBash
ModeCLI infra
Evidence mediaScreenshot + artifact telemetry
MAINFRAME screenshot
A visual proof surface for a bash-native standard library built around agent infrastructure.
Artifact typeCLI infrastructure proof

Highlights authorship, systems range, and inspectable open-source engineering.

Open motion artifact ->
4,000+ functionsPure bash runtimeAgent primitives
Artifact telemetryInspection trace
Capture classSystem artifact
Proof posturePublic repo
System signalBash primitives
Problem

Agentic shell workflows are usually a mess of brittle scripts and duplicated glue logic. They need durable primitives, not one-off hacks.

Architecture map
01Pure bash runtime
02Agent working memory primitives
03CLI-first orchestration surface
Build decisions
  1. Built a broad bash-native standard library instead of relying on heavyweight runtime assumptions.
  2. Added memory, orchestration, and data-handling primitives so shell workflows could scale into systems.
  3. Framed the project as infrastructure for AI-native tooling, not just a utilities repo.
Proof signals
  1. Strongest evidence of systems-level engineering range in the portfolio.
  2. Gives technical peers a concrete artifact to inspect, evaluate, and trust.
  3. Signals authorship and initiative beyond client delivery work.
Outcomes
  1. Signals systems-level engineering range
  2. Shows open-source authorship
  3. Gives technical peers something concrete to inspect
PropertyJunk.com
052024Live

PropertyJunk.com

Junk removal SaaS with integrated CRM

Impact lineOperational complexity turned into a single coherent software surface.
ModeOps SaaS
SurfaceCRM + fleet + billing
FocusWorkflow continuity
System topologyInput -> core -> outcome
01
Input surfaceJobs + field signals

Customers, employees, schedules, fleet movement, billing state, and service requests enter one operating model.

02
System coreOps coordination layer

CRM, scheduling, fleet visibility, and billing workflows are treated as one product surface.

03
Output layerDispatch-ready SaaS

Staff can coordinate real-world work without switching between disconnected tools.

Control signals
Job lifecycle stateFleet visibilityRevenue workflow continuity
Proof telemetry
SurfaceOps SaaS
ScopeCRM + fleet
WorkflowDispatch
Evidence mediaScreenshot + artifact telemetry
PropertyJunk.com screenshot
A field-operations product surface for scheduling, CRM, fleet visibility, and revenue workflow continuity.
Artifact typeDispatch software UI

Shows Gordon can design around real-world operational movement and not just static screens.

Open motion artifact ->
CRM + schedulingFleet operationsRevenue workflow
Artifact telemetryInspection trace
Capture classOps workspace
Proof postureProduct surface
System signalCRM + fleet
Problem

A field-operations business needed software that could coordinate jobs, people, vehicles, customers, and billing without forcing staff to bounce between disconnected tools.

Architecture map
01Operations dashboard
02Fleet and scheduling workflows
03Unified CRM + billing
Build decisions
  1. Treated operations as a product surface, not just a CRUD admin panel.
  2. Designed around scheduling, fleet visibility, CRM, and financial workflow continuity.
  3. Used frontend structure to reduce operational friction for a real-world service business.
Proof signals
  1. Shows range across dashboards, workflows, and operational software design.
  2. Demonstrates practical SaaS thinking tied to actual business movement.
  3. Signals competence with systems that touch revenue-critical operations.
Outcomes
  1. Turns field operations into software
  2. Demonstrates breadth across product surfaces
  3. Shows real-world operational complexity
toddwatts.dev
062026In progress

toddwatts.dev

This site — a living experiment

Impact lineA portfolio that increasingly behaves like a product demo for the person behind it.
StackNext.js 16 + STDB
ContentAPI-publishable
ModeLive experiment
System topologyInput -> core -> outcome
01
Input surfaceContent + signal inputs

MDX posts, external feeds, project metadata, publish requests, and ops events enter one site system.

02
System coreNext.js intelligence layer

Aggregation, ranking, route handlers, browser gates, and ops history turn the portfolio into software.

03
Output layerPublic proof surface

Visitors see live signals, case files, operational telemetry, and engineering evidence in one interface.

Control signals
Bearer publish authSource diversity capsOps history retention
Proof telemetry
RoutesAPI + UI
QualityBrowser gated
HistoryBlob-ready
Evidence mediaScreenshot + artifact telemetry
toddwatts.dev screenshot
Generated Ghost Signal media and live operational layers make the portfolio itself part of the proof.
Artifact typeGenerated social card

Connects visual identity, metadata, content routes, and production verification into one system.

Open motion artifact ->
Generated OG routesOps consoleBrowser gates
Artifact telemetryInspection trace
Capture classGenerated route
Proof postureBrowser gated
System signalOpenAPI + ops
Problem

A portfolio should not behave like a static brochure if the goal is to signal systems thinking, frontend ambition, and live technical credibility.

Architecture map
01Signal feed aggregation
02Headless publish API
03Realtime-ready portfolio shell
Build decisions
  1. Turned the site into a product surface with publishing, signal aggregation, and reusable data layers.
  2. Used the portfolio itself as a proving ground for design language and information architecture.
  3. Continues evolving toward a live informational engineering showcase instead of a generic personal site.
Proof signals
  1. The site is becoming a public artifact of how content, systems, and presentation connect.
  2. Demonstrates willingness to iterate based on critique instead of treating the portfolio as finished.
  3. Makes the frontend itself part of the proof surface.
Outcomes
  1. Acts as a public product demo
  2. Shows how content, design, and systems connect
  3. Lets the site itself become proof of capability
Pi CMUX
072026Live

Pi CMUX

Local multi-agent command center for Pi + CMUX

Impact lineTurns local Pi + CMUX work into an agent command center with shared state and recoverable handoffs.
Extensions4-part family
ModeLocal-first
SurfaceTerminal + browser
System topologyInput -> core -> outcome
01
Input surfaceLocal agent sessions

Pi workflows, CMUX workspaces, browser observations, locks, checkpoints, and session state are bridged.

02
System coreCoordination substrate

Extensions coordinate terminal and browser work with recoverable state and shared operational memory.

03
Output layerAgent command center

Local multi-agent teams can operate with handoffs, status, and recovery instead of scattered context.

Control signals
Session bridgesBrowser locksCheckpoint recovery
Proof telemetry
Extensions4-part
ModeLocal-first
SurfaceCMUX + browser
Evidence mediaScreenshot + artifact telemetry
Pi CMUX screenshot
A local-first multi-agent workflow surface framed around Pi, CMUX, browser checkpoints, and recoverable handoffs.
Artifact typeOpen-source package proof

Uses a public repository and installable extension family as the inspected artifact.

4-part extension familySession bridgeBrowser checkpoints
Artifact telemetryInspection trace
Capture classPackage artifact
Proof posturePublic repo
System signalCMUX bridges
Problem

Local agent work often loses shared context across terminals, browser surfaces, handoffs, and long-running team coordination.

Architecture map
01Agent orchestration layer
02Pi session bridge
03Browser lock + checkpoint system
Build decisions
  1. Packaged four coordinated Pi extensions as one installable family for CMUX-powered workflows.
  2. Added bridges between Pi sessions and CMUX workspaces, surfaces, events, and status records.
  3. Built browser intelligence primitives for locks, checkpoints, observations, recovery, and durable memory.
Proof signals
  1. Public open-source repo with installable Pi package metadata and detailed documentation.
  2. Shows local-first multi-agent orchestration across terminal and browser surfaces.
  3. Connects agent coordination, browser automation, and design workflows into one operating layer.
Outcomes
  1. Makes local agent teams easier to coordinate
  2. Connects browser and terminal work surfaces
  3. Creates reusable open-source agent infrastructure