Ethan Marsh is the engineer who makes the pieces talk. He designs and builds the agentic workflows, API integrations, and data orchestration layers that sit on top of on-vessel AI infrastructure — the software that turns expensive hardware into guest concierges, L1 support agents, voice assistants, and offline knowledge tools.
His background blends product management with hands-on AI engineering. Before ShipboardAI, he worked on enterprise AI search and recommendation systems, integrating large models with CRMs, telephony stacks, legacy databases, and marketing automation. Carnegie Mellon MBA. Strong opinions about when to use an agent and when to use a function call.
Ethan writes about the practical side of local AI deployment: how to wire a 70B model into a ship's PMS without latency spikes, how to run multi-agent workflows that degrade gracefully when one tool goes offline, how to ship voice assistants that work in seven languages on a vessel with no internet. The nuts and bolts of making on-prem AI feel like magic.