Skip to content
@sawfwair

Sawfwair Inc.

Founder-led technical practice building local-first AI infrastructure and providing fractional CTO leadership.

Sawfwair — Software systems that stay clear as you grow

Sawfwair

Founder-led technical leadership for consequential software, architecture, AI, and delivery decisions.

Sawfwair is a software practice based in Prince Edward Island and working globally. We embed with growing teams that need the judgment of a senior technical leader and the output of a hands-on builder — without pretending strategy and delivery are separate jobs.

Founder Kyle McCullough brings a career spanning film production, games shipped with Ubisoft, multi-million-dollar U.S. Department of Defense research at USC, enterprise AI, and CTO leadership. Sawfwair turns that range into clear decisions, durable systems, and shipped products.

What we do

  • Fractional CTO leadership — retained technical ownership across product, architecture, teams, vendors, and delivery.
  • Critical product delivery — embedded 6–12 week engagements when an important system has to move.
  • AI systems — grounded model, agent, retrieval, computer-vision, and workflow infrastructure beyond chatbot wrappers.
  • Architecture and performance — designs that survive real users, real constraints, and the next stage of growth.
  • Agent-accelerated execution — coordinated AI workflows with human accountability and proof at every boundary.

What we are building

Mere — create anything, locally

Mere is a local-first AI ecosystem: one runtime contract from a single machine to private remote compute, visual workflow authoring, plugins, and agent-ready workspace operations.

Surface Repository Role in the system
Runtime mere.run Local multimodal inference, model management, training, workflows, and API serving.
Studio Mere Graph Studio Portable visual workflow authoring and operations.
Compute Mere Relay Private remote scheduling and outbound-only compute nodes.
Operations MereKit Onboarding, context, diagnostics, MCP tools, and workspace operations.
Connections MereKit Link Declarative links, sync policy, and integration boundaries.
Extensions Mere plugins User-controlled bridges to external compute and providers.

The public repositories include runnable software, release artifacts, architecture decisions, operational boundaries, tests, and documentation. The aim is not to talk about local-first AI in the abstract; it is to make the claim inspectable.

Work with Sawfwair

Need senior technical leadership for a consequential product, architecture, AI, or delivery decision?

Start a conversation →
sawfwair.com · kyle@sawfwair.com · Prince Edward Island, Canada

Popular repositories Loading

  1. mere-run mere-run Public

    Run local image, text, speech, vision, music, and video workflows, plus model management and a loopback OpenAI-compatible API server. The repo ships the core libraries and the mere.run executable.

    Swift 52 2

  2. merekit-cli merekit-cli Public

    One command plane for Mere workspaces: TUI onboarding, manifest-backed app CLIs, read-first ops, agent context packs, and MCP tools.

    JavaScript 2

  3. dkcli dkcli Public

    Proof-driven design CLI for OKLCH palettes, APCA contrast, fluid scales, motion curves, layout audits, tokens, and Svelte components.

    TypeScript 1

  4. merekit-link merekit-link Public

    MereKit CLI for declaring workspace links, sync policy, and Executor-backed integrations.

    TypeScript 1

  5. relay-mere-run relay-mere-run Public

    Account-scoped relay and desktop compute node for private, local mere.run inference.

    TypeScript 1

  6. mere-run-plugins mere-run-plugins Public

    Official companion plugin contracts and provider runners for mere.run.

    Python

Repositories

Showing 10 of 11 repositories

Top languages

Loading…

Most used topics

Loading…