r2n.io/Ryan McDonough

Workflows forAI-era teams.

Ryan McDonough, AI software engineer. I cultivate the workflows that let software teams build well with agents: Claude Code rollouts, MCP servers that give agents governed access to real systems, shared prompt libraries, and review habits that keep engineers owning what ships. Ten years of production TypeScript, Java, and C# behind it.

Email Ryan Resume PDF
Startup TBAPrincipal Software ArchitectSep 2026 – Present
Wait for hundreds of loan officers to load loan data
3–5 min → 3 sec
merged PRs a week at my current startup
170+
services moved onto one CI/CD pattern
80+ Pre-AI
paying users on the credit products I led
Millions

Workflows I've built for teams

Each one changed how a team ships: a governed AI platform, an agent toolchain, a live SDK migration, an event pipeline, a testing habit, and a CI/CD platform.

  • Claude Code
  • MCP servers
  • Agentic workflows
  • Prompt + tool design
  • TypeScript
  • C# / .NET
  • React / Next.js
  • Event-driven systems
  • Terraform

AI a regulated small business can trust

1,566merged PRs in nine weeks

  1. Desktop client
  2. Governed AI services
  3. Durable workflows
  4. Reviewed actions + receipts
Constraint
Small and mid-size businesses in regulated industries want AI help with their mail, files, and plans, but can't hand an assistant their business data without knowing what it reads and what it changes.
What I built
I took a vibe-coded .NET prototype to a production platform, desktop client to Azure, and built the governance into it: per-connection controls over what the assistant and outside agents read, reviewable batches for every AI action, verified privacy erasure, and OpenTelemetry with answer ratings feeding PostHog. Coding agents shipped it with me under a second-model review and proof on the real app.
What changed
1,566 merged PRs across 43 repositories in nine weeks, a peak of 417 in one week, and a product that measures its AI's answers instead of guessing.
PRs shippedMy contributionsto AI tooling

TypeScript / Electron / Python / Azure Container Apps / OpenTofu / PostgreSQL / OpenTelemetry

Claude, wired into the loan system

  1. Encompass APIs + legacy SDK
  2. MCP server
  3. Claude Code + Copilot
  4. Shared prompt libraries
Constraint
Loan-platform knowledge lived in a legacy .NET SDK and in people's heads. Agents couldn't see the systems engineers work on, and regulated lending rules out copy-paste AI.
What I built
I built a production MCP server that gives Claude governed access to Encompass across the Developer Connect APIs and the legacy SDK. Then I rolled Claude Code and GitHub Copilot out across the engineering org with shared prompt libraries and custom MCP tooling.
What changed
Review and scaffolding got faster. Agents took the repetitive work; architecture and production ownership stayed with engineers. The pilot became daily infrastructure.

MCP / Claude Code / GitHub Copilot / TypeScript / .NET / Encompass

Off a dying SDK, live

0 downtimelive SDK/API migration

  1. Legacy SDK
  2. Developer Connect APIs
  3. Azure Container Apps
  4. Custom workflow UX
Constraint
ICE set an end-of-life date for the .NET SDK every Encompass integration ran on. Workflows had to keep running through the swap.
What I built
Rebuilt the integrations on Developer Connect APIs as containerized TypeScript and C# services on Azure Container Apps, deployed with Terraform. Where the APIs lacked parity, like the SDK's popups, we built web UX to match.
What changed
A live swap with no downtime on a vendor's timeline, and an integration layer that now ships independently of workflow changes.

TypeScript / Azure Container Apps / Terraform / Docker / C# / Encompass

Minutes of waiting, down to seconds

3–5 min → 3 secwait to load loan data

  1. Webhook ingestion
  2. Event Grid
  3. Container Apps
  4. Async downstream jobs
Constraint
Hundreds of loan officers waited 3–5 minutes for their software to load loan data, because every step ran on their own machines, one after another. The business tracked load times daily.
What I built
Webhook ingestion and Azure Event Grid pipelines that move email, notifications, and downstream jobs into async flows, off the client.
What changed
The wait fell to 3 seconds, and teams started asking to move more work onto the pipeline.

TypeScript / Azure Event Grid / Azure Container Apps / Webhooks / Async processing

A credit app for millions

Millionspaying consumer users

  1. Product delivery
  2. React/Next.js
  3. Test standards
  4. Continuous releases
Constraint
A credit-monitoring dashboard for millions of paying users needed continuous features (BNPL, personalized video, UX refreshes) during a platform modernization.
What I built
Led teams of 10+ through multiple redesigns without pausing delivery, and made unit testing a company-wide practice, training dozens of teams on their first tests.
What changed
Moved the frontend to React, Next.js, and Vite, and grew engineers into senior roles. The testing habit outlasted any feature.

TypeScript / Java / React / Next.js / Vite / Static analysis

One pipeline, 80+ services

80+services on one delivery pattern

  1. Source commit
  2. Jenkins
  3. Quality gates
  4. Helm/Kubernetes
Constraint
Every deploy meant SSH, curling WAR files into place, restarting WebLogic, and configuring servers by hand.
What I built
With two other engineers: Jenkins pipelines, Helm and Kubernetes deploys, Terraform, and SonarQube gates. Running within a month, then refined as teams onboarded.
What changed
80+ service teams got repeatable builds and deploys. Ending the manual deploy ritual changed how they shipped more than any gate did.

Kubernetes / Helm / Terraform / Jenkins / Docker / SonarQube

Ten years in production

  1. Sep 2026 – Present

    Startup TBAPrincipal Software Architect

    Unannounced company. I joined to turn a vibe-coded .NET prototype into a production platform, and led the v2 relaunch that made the desktop installer the main product.

    • Took the prototype to production scale: rebuilt it in TypeScript on Effect, traced the services and desktop client with OpenTelemetry into PostHog, and moved long-running AI work onto durable workflows that survive restarts.
    • Led the v2 relaunch around the Windows installer and Electron desktop app, backed by Python services on Azure Container Apps behind API Management, PostgreSQL, and the OpenTofu that runs it all.
    • Built governance into the product, not around it: admins choose what the assistant and outside agents like Claude and ChatGPT can read per connection, AI actions land as reviewable batches with receipts, and privacy erasure is a durable, verified job.
    • Merged 1,566 pull requests across 43 repositories in my first nine weeks, averaging 170+ a week and peaking at 417, with coding agents in the loop under rules I wrote.
    • Built the pipeline those agents ship through: durable per-worktree environments, a faster CI, a second-model review gate, verify skills that drive the real app before a PR is ready, and mutation testing to check the tests themselves.

    TypeScript / Effect / Electron / Python / Azure / OpenTofu / PostgreSQL / OpenTelemetry

  2. Oct 2024 – Aug 2026

    Summit Funding, Inc.Lead Systems Architect

    AI-native developer tooling and event-driven loan integrations: MCP servers, Encompass, Azure Event Grid, TypeScript, C#.

    • Built a production MCP server integrating Claude with the Encompass loan origination platform, bridging modern Developer Connect APIs and legacy .NET Framework SDK calls.
    • Rolled out Claude Code and GitHub Copilot across the engineering org with shared prompt libraries and custom MCP tooling — accelerating code review and feature scaffolding without replacing architecture review or production ownership.
    • Rebuilt legacy Encompass SDK integrations around Developer Connect APIs and containerized TypeScript and C# services in Azure Container Apps with Terraform-managed deployments.
    • Designed and owned the webhook and Azure Event Grid pipelines that cut the wait for hundreds of loan officers to load loan data from 3–5 minutes to 3 seconds.
    • Standardized releases and service design using Terraform, Bitbucket Pipelines, and C4 architecture diagrams across the engineering org.

    TypeScript / MCP / Claude Code / Azure Event Grid / Azure Container Apps / Terraform / C#

  3. Nov 2021 – Sep 2024

    TransUnion Consumer InteractiveLead Full Stack EngineerPre-AI

    Consumer credit products serving millions of users: TypeScript, React, Next.js, Java, Python.

    TypeScript / React / Next.js / Vite / Java / Python

  4. May 2021 – Nov 2021

    TransUnionLead Engineer, CI/CD PlatformPre-AI

    DevOps and developer experience: Kubernetes, Helm, Terraform, Jenkins, SonarQube.

    Kubernetes / Helm / Terraform / Jenkins / Docker / SonarQube

  5. Feb 2018 – May 2021

    TransUnionSenior Engineer, ETL & PlatformPre-AI

    Enterprise Java services and data pipelines: Ab Initio, JUnit, Selenium, Splunk, Grafana.

  6. Sep 2016 – Feb 2018

    TransUnionAssociate QA EngineerPre-AI

    QA on consumer credit features including MultiBureau credit lock and Ab Initio batch processing.

  7. Nov 2014 – Sep 2016

    HathwayQA Analyst & InternPre-AI

    QA across enterprise client accounts at a digital agency later acquired for nine figures.

Putting agents into how your team builds?

I'm looking for AI software engineering work: product teams writing real software with agents in the loop. Email is fastest; the resume has every date.

Ryan@r2n.io
  • Greater San Luis Obispo Area
  • Remote preferred
  • Open to relocation