Portfolio / Systems architecture

Six products.
One systems lens.

These case studies show how I frame domain boundaries, connect tools to their execution environments, and make the result understandable to the person using it. Each product has its own architecture; the shared pattern is a design approach, not a shared codebase.

4 App Store products 1 app in internal testing 1 playable prototype

Cross-project system design

Keep the handoffs visible.

Across AI workspaces, music tools, and simulations, I use the same portfolio lens: clarify what the user wants, keep domain decisions in their own workflow, make external execution explicit, and return a result the user can inspect.

  1. 01IntentWhat is the person trying to do?
  2. 02Domain workflowWhich state and decisions belong to this product?
  3. 03Execution boundaryWhich provider, audio engine, or simulation acts?
  4. 04Result & evidenceWhat changed, and how can it be checked?

Problem · architecture · integrations · tradeoff · evidence

The work, system by system

Release labels describe public product state. Candidate builds and unverified outcomes stay out of release claims.

01 / Local AI orchestration

Internal testing

ComfyChat / Conductor

A mobile cockpit for local model and image workflows.

Problem

Make chat and image-generation work manageable when models and engines can run on different devices.

Architecture

A native iPhone and iPad client organizes Chat, Create, Models, and Settings. Local and on-device models and ComfyUI image jobs remain distinct execution targets, with the mobile app providing the interface for configuring and starting work.

Integrations & boundaries

Connects to local AI services and ComfyUI workflows. The public product page describes the product as being in internal testing.

Tradeoff

Supporting different runtimes expands choice while increasing setup and readiness work. The interface needs to make the selected model and execution target clear.

Verified outcome

The public project page identifies ComfyChat as being in internal testing. No performance figures are claimed.

Current product state

Internal testing. Public App Store release is not claimed.

02 / Model workspace

App Store

LM Manager Pro

“An agentic harness in the palm of your hands.”

A local-first creation studio spanning different model providers.

Problem

Bring model setup, creative work, and generated assets into a mobile workspace for local, on-device, and cloud models.

Architecture

A local-first mobile workspace routes creation through explicit provider choices. Chat, image creation, reusable personas and teams, Library, and Activity make up the visible product workflow.

Integrations & boundaries

Supports local model services, on-device models, and cloud providers. Model choice is explicit; the portfolio does not claim that the app autonomously takes external actions.

Tradeoff

Provider breadth adds capability and configuration overhead. Keeping routing, creation, and saved work legible helps the user understand where a task runs and what it produced.

Verified outcome

The public App Store listing identifies LM Manager Pro as a released product. No performance or adoption metrics are claimed.

Current product state

App Store release.

03 / Aerospace strategy simulation

App Store

Fabricon: Aerospace

An aviation-industry company simulation across changing eras.

Problem

Make the growth of an aerospace company playable across aircraft programs, people, research, contracts, finances, and historical change.

Architecture

The game organizes a company simulation around historical eras and connected business systems. Campus operations and aircraft programs lead into Flight Test, a simulated in-game activity.

Integrations & boundaries

The described product connects its strategy and flight-test systems inside the game. Flight Test is a simulation; no real aircraft, flight data, or operational control is claimed.

Tradeoff

More management depth creates more navigation and state to explain. A company-and-program framing keeps workforce, research, finance, contracts, and aircraft work inside one understandable game world.

Verified outcome

The public App Store listing identifies Fabricon: Aerospace as a released strategy game. No real-world aerospace outcome is claimed.

Current product state

App Store release. Flight Test is an in-game simulation.

04 / Mobile audio workflow

App Store

PocketDAW: Mobile Studio

A local-first vocal studio that carries a song from capture to export.

Problem

Give a musician a usable recording and production workflow on iPhone and iPad without splitting the song across disconnected tasks.

Architecture

A music project is the organizing unit across recording, arranging, editing, mixing, lyrics, and export. Playback and rendered output stay within the same production workflow.

Integrations & boundaries

Uses Apple-platform audio input and playback, with portable export. An optional user-operated LM Studio connection supports assistant features; core music work stays within the project workflow.

Tradeoff

Precision controls compete with limited screen space and touch input. Timeline zoom, track sizing, and focused trimming help keep edits manageable on mobile.

Verified outcome

The public App Store listing identifies PocketDAW as a released vocal-production studio. No audio-quality or creator-output metric is claimed.

Current product state

App Store release.

05 / Player-controlled tuning

App Store

ForzAdvisor

A companion workflow for reviewing and testing tune changes.

Problem

Help a player turn vehicle details and tuning goals into a set of changes they can review and try in the game.

Architecture

The implemented flow is user-led: supply car details by entry or photo, inspect a tune plan, then apply supported settings in game and evaluate the result. The player owns the final test loop.

Integrations & boundaries

Works from user-supplied car details. FH6 is the numeric-tuning path; FH5 remains a build-plan and research path. No bundled car roster, live telemetry, or in-game settings writeback is claimed.

Tradeoff

Manual confirmation keeps the player in control and works without relying on an unsupported game-control interface, while making the real in-game drive necessary to validate each change.

Verified outcome

The public App Store listing identifies ForzAdvisor as a released companion. This case study makes no accuracy, lap-time, or handling-improvement claim.

Current product state

App Store release. The player and game provide the validation feedback; the companion does not claim to write settings into the game.

06 / Civic and political simulation

Playable prototype

The Long Republic

A mobile-first political dynasty prototype grounded in local civic decisions.

Problem

Make political change understandable through the daily decisions, people, and institutions of a place.

Architecture

The prototype begins with a founding settlement and builds toward a long political time horizon. Residents, assignments, petitions, and infrastructure support the larger dynasty and regime-change arc.

Integrations & boundaries

The current product is a contained Unreal prototype. Simulation and character content are in-game systems; no external political-data integration is claimed.

Tradeoff

The first chapter must teach decisions clearly while leaving room for consequence over a longer horizon. Character expansion remains in development; no total is claimed.

Verified outcome

The public project page identifies The Long Republic as a playable prototype. No completed roster count or public release is claimed.

Current product state

Playable prototype. Founding and frontier content remain in development.

Evidence standard

Say what is live. Label what is still being built.

Public release pages, internal candidates, and prototypes are different product states. This portfolio names a state only when a public listing or project page supports it; no usage counts, performance benchmarks, or third-party endorsements are presented.