Schwen Scalability
Menu
Schwen Scalability

Founded by Matthew Schwen in Franklin, TN and Lehi, UT.

AI at real scale.

I build LLM systems and agents that don’t fall over, with twenty years of running production behind me.

Matthew Schwen
Matthew SchwenSr. Site Reliability Developer at Oracle Cloud. Founder of Schwen Scalability.
From pulling fiber in 2006 to running exabyte-scale storage at Oracle today.Featured in Network World. Keynote at the ONUG AI Networking Summit. Guest on Packet Pushers.Shipped and runningChekkd, RnBy, CarrierOS, CHESS, SaltBot, LaunchWatch

From the fiber up.

More than twenty years on call for every layer an AI system runs on: the fiber, the data center, the network, the platform, and now the AI on top.

  1. 2004to 2008

    VCTEKM

    The fiber and the NOC

    Veracity Communications, TEKsystems, Mstar. Provo, UT

    Started in a network operations center provisioning T1s, DSL and business phone systems, pulled fiber into new stores, then ran the core network for UTOPIA’s fiber-to-the-home service: data, voice and TV for more than 20,000 homes, on call around the clock.

  2. 2008to 2015

    Adobe

    Data centers

    Adobe. Lehi, UT

    Seven years building Adobe’s own data-center routing, switching and firewalls, its Lehi campus network, fully meshed and non-blocking, and the networking inside and around its OpenStack cloud. On call for the network worldwide.

  3. 2015to 2017

    SymantecAlianza

    Global operations

    Symantec, then Alianza. Lindon, UT

    Led engineering for Symantec’s global network operations: the escalation point for Severity One outages, root-cause analysis, and the routing standards that followed. At Alianza, a cloud voice platform for service providers, rebuilt the Ashburn data center on an Ansible-managed spine and leaf.

  4. 2017to 2019

    Adobe

    The network as software

    Adobe. Lehi, UT

    Made the network self-service: one API call created a VLAN, reserved its subnet and DNS, and configured the switches and firewalls. Built Adobe’s first 100G backbone between data centers, a transit hub into AWS, and an auditor that finds configuration errors on its own.

  5. 2019to 2023

    The Trade Desk

    Reliability at ad-tech scale

    The Trade Desk, Staff Software Engineer. Remote

    Network reliability for an advertising platform answering billions of queries a day: a fully automated Cumulus and eBGP network across the WAN, AWS and Azure, Go-based BGP load balancers, and an automation platform on Kubernetes used across the company. Promoted to Staff, and hired the team.

  6. 2021to 2025

    PortalCX

    Founding a company

    PortalCX, co-founder and CTO. Lehi, UT

    Co-founded a customer-onboarding platform with my brother and a close friend, and wore every hat for four years: engineering, design, recruiting, fundraising and marketing.

  7. 2024to 2025

    Humana

    Automation and AI in the enterprise

    Humana, Associate Director of Network Automation Engineering

    Set Humana’s network-automation strategy and led its senior engineers across hundreds of vendors and regions: pipelines through Azure DevOps, ServiceNow and Ansible, a source-of-truth platform, and an AI co-pilot for network operations, unveiled in my ONUG keynote with Network to Code.

  8. 2025to now

    OracleChekkdSchwen Scalability

    Exabyte storage and AI products

    Oracle, Chekkd, Schwen Scalability

    Sr. Site Reliability Developer for Oracle Cloud Infrastructure Object Storage, at exabyte scale. Co-founder and CTO of Chekkd. And this practice: AI systems for founders, built on all of the above.

Computer science, University of Phoenix. Full résumé on request. Employers and roles held, not clients of Schwen Scalability. Company names and marks belong to their owners and are shown to describe where the work was done; they do not imply any endorsement, sponsorship, or affiliation.

Work

Products I founded, built for clients, or engineer inside as a technical partner. Every frame is a capture from the live site.

Chekkd (opens chekkd.com in a new tab)

Trust platformCo-founder & CTO

Identity verification for peer-to-peer deals: a four-level trust ladder with Plaid-backed ID checks, biometric liveness, and a portable TrustBadge.

RnBy (opens rnby.ai in a new tab)

AI agentsCo-founder, product and engineering

An AI operating layer for small business. Run your company from a conversation, with multi-agent routing, shared memory, and cited answers.

CarrierOS (opens carrieros.ai in a new tab)

AI infrastructureTechnical partner, core systems

Inference control system that stabilizes long-running AI work: state-aware execution, persistent memory, multi-model orchestration.

CHESS (opens chess-if.com in a new tab)

Legal SaaSBuilt end to end, run as a SaaS

Case management for advocacy organizations: workflow state machines, a secure family portal, HIPAA-minded audit logging.

More builds

Open source

  • Claw Proxy (opens GitHub in a new tab)MIT, TypeScript. mattschwen/claude-max-api-proxy

    OpenAI-shaped gateway into your authenticated Claude Code CLI: dynamic model routing, session resume, zero extra API key.

  • Synapse (opens GitHub in a new tab)MIT, Python. mattschwen/Synapse

    Autonomous AI agent for SRE and Ops on Semantic Kernel: dynamic incident-response planning, native code execution, vector memory, self-learning function reuse.

In The Beginning

A new AI product, in stealth

Founded and building now, heads-down. Details over a call; happy to demo under NDA.

Ask me about it

More under NDA, with references on request. Everything open source is at github.com/mattschwen.

Services

Every engagement is scoped the way an SRE plans a migration: explicit, measurable, reversible.

Under each service: where that kind of work was done. Employers and ventures, not clients.

  • AI Systems & ModelsComplex AI that survives production.

    How I build complex AI that survives production: multi-agent orchestration with shared memory, RAG with cited answers, knowledge graphs over messy documents, MCP tool integrations, and evals that tell the truth about whether any of it works.

    You get agents and orchestration, RAG and knowledge graphs, MCP and tool integrations, evals and guardrails. Done at RnBy, and Synapse (open source).

  • Site Reliability EngineeringKeeping systems boring since 2004.

    Twenty years of keeping systems boring, from fiber NOCs to exabyte-scale Object Storage at Oracle. SLOs that mean something, incident response that finds root cause, automation that removes the 3am page instead of snoozing it. Reliability isn’t a feature; it’s the product your customers never have to think about.

    You get exabyte-scale operations, SLOs and incident response, blast-radius design, Sev-1 escalation leadership. Done at Oracle, The Trade Desk, Symantec.

  • Infrastructure at Scale, using AIThe Humana playbook: infrastructure that runs itself.

    AI-assisted infrastructure operations, the playbook that made the trade press: turn workflow automations into prompts, wire a source of truth, and let systems self-serve. Networks, cloud, and data centers that run themselves on rails.

    You get AI-assisted automation pipelines, source-of-truth platforms, self-service infrastructure. Done at Humana, Adobe, Alianza.

  • ConsultingSenior technical judgment, on demand.

    Architecture reviews, AI strategy, build-vs-buy calls, and the hard conversations about what your system actually needs, from someone who has to live with the consequences of decisions like yours every day.

    You get architecture and AI strategy, build-vs-buy decisions, hands-on work, not slideware. Done at SaltBot, CHESS.

  • PartnershipsInside your team, carrying the pager with you.

    Embedded technical partner or fractional CTO, the CarrierOS model. I sit inside your team, own core systems engineering, and carry the pager with you.

    You get a fractional CTO, core systems ownership, a long-horizon engagement. Done at CarrierOS, Chekkd, PortalCX.

  • Start-upsZero to one, one to scale, and the diligence in between.

    Zero-to-one and one-to-scale: MVPs shipped in weeks on the standard stack, growth architecture when it works, and technical due diligence for the VCs writing the check.

    You get MVP to production, scaling architecture, technical due diligence for VCs. Done at Chekkd, RnBy, LaunchWatch.

Who you’ll work with

Systems don’t fail at the pieces. They fail where the pieces meet, so the joints are what I engineer first.

Where the auth meets the database. Where the agent meets the tool. Where the demo meets production. Twenty years of keeping systems alive, from fiber NOCs to exabyte-scale cloud storage, taught me that lesson on repeat, and Schwen Scalability is where I apply it for founders and operators.

The AI work isn’t prompt tinkering: retrieval with cited answers, multi-agent orchestration with shared memory, knowledge graphs over messy real-world documents, and evals that tell the truth. I bring an SRE’s paranoia to AI: measure it, cap its blast radius, make it reliable enough to bill for.

Snow on Mount Timpanogos at dusk, above the lights of a Utah Valley town

I lead every engagement personally, with a bench of senior developers and specialists behind me. The person who architects your system still writes the code, runs the deploy, and answers the phone.

Mount Timpanogos over Utah Valley, near the Lehi base.

Two home bases, and on the road a lot.

Lehi, UTFranklin, TN
Home bases
Franklin, TN and Lehi, UT
Local time
Central time in Franklin, Mountain time in Lehi
In production since
2004
Right now
Sr. Site Reliability Developer, Oracle Cloud

Let’s make the pieces fit.

Tell the intake agent what you’re building. It verifies your email, reads your docs, and requests an intro call with Matthew.

Prefer email? matthew@schwen.me