Norfolk Southern · Loco Works
01 / 08Mechanical Supervisor, Altoona Locomotive Works
SIGNAL ACQUIRED // PRINCIPAL SOFTWARE ENGINEER · AI-NATIVE DEVELOPER WORKFLOWS · CUSTOMER-FACING ENGINEERING
Principal Software Engineer at Comcast — AI Technical Lead on the X1 platform, 10+ years shipping production software. I build AI-native developer workflows and the customer-facing systems they serve. Before this: locomotives, Navy ships, a menu-board startup, Fortune 500 consulting, robot safety controllers, smart TVs. This page is an instrument, and right now it’s measuring you. Check the strip at the top.
BEGIN SWEEP ▼
Eight events on a single line. Keep scrolling to scrub the cursor through time — each spike is a place where the signal changed. The last one is still transmitting.
Mechanical Supervisor, Altoona Locomotive Works
Every panel below is a real system with real users. Channel colors: ■ frontend ■ backend ■ infra ■ ai.
Solo-built production platform: design your characters, get AI-illustrated stories, order them as real hardcover books. I own every layer — prompt iteration, cost/latency tradeoffs, blue/green rollouts, E2E testing, and full production observability: tracing, metrics, alerting. A telemetry page respects a system that reports its own vitals.
Multi-tenant household finance, live on the App Store and Google Play — one budget, every phone in the house sees the same numbers — with AI spending coaching, performance charting, cloud functions, and push notifications. Publishing to both stores is automated through GitHub Actions: tag, build, ship.
Cross-platform goal tracker with an AI coach built in — check-ins and plan generation driven by your health data, your goals, and how you’re actually performing. One Flutter codebase, four kinds of screen.
A locally hosted MCP server feeding custom tools to my AI agents — instrumentation for the instruments. On a telemetry page, this is the panel that matters: the plumbing that lets an agent read the world. Example tool: a resilient weather cache that keeps answering when the upstream API won’t.
Forward-deployed engagement with Accenture’s first entrepreneur-in-residence startup — healthcare AI. I architected the core product: a pipeline ingesting wearable healthcare data into their AI engine, the ETL behind it, and the B2B and B2C portals in front of it.
JIRA for contractors: bidding and estimation tracking — data entry, dashboards, reports, charting, exports, file attachments. Websockets keep every open dashboard current, so the office and the job site read the same numbers.
DECOMMISSIONED / STANDBY SYSTEMS · 2018–2025
Same engineer on every channel. That’s the point — I’ve owned whole systems alone, so I know what each layer costs the others.
Where I live day to day — I’ve led UI teams from telemetry dashboards to smart-TV apps to robot-fleet portals, so state management and render performance are the job, not a nice-to-have. NgRx re-architectures and D3 instrument panels are home turf.
APIs, data models, and the integrations that move money and mail — from Stripe checkouts real businesses depend on to the mission-critical Java migration I’m steering right now.
From my own rack to two clouds: containers, orchestration, reverse proxies, TLS — plus the pipelines that ship it all, and the rollout patterns (blue/green, canary) that keep shipping boring.
The channel the whole board watches now. At Comcast I lead AI-native developer workflows adopted company-wide — vulnerability-fix agents, AI-assisted migrations, generated tests. At home, the same discipline ships myInkwell and taskr. I integrate models the way I integrate any unreliable upstream: carefully, with fallbacks and a human in the loop.
FORMAL CALIBRATION · MOST RECENT FIRST
If you’re building something worth monitoring — or you just want to argue about RxJS operators — the channel is open.
$contact --via email
>software@ryanwilk.com
$contact --via github
>github.com/rjw1428
$contact --via linkedin
>linkedin.com/in/rjw1428
$contact --via web
>ryanwilk.com