~/projects $ cat projects/dawn.md
Dawn
activeproprietaryResMed · San Diego, CA
ResMed's AI-powered digital sleep-health assistant. I was one of the engineers who built it.
Dawn is ResMed's generative-AI digital assistant for sleep health. I contributed to building Dawn as part of my machine-learning engineering work at ResMed. Public information about the product is described on ResMed's own channels. My personal contributions are described here only at the level that is public-safe.
try it
Dawn is available to myAir users in the app and on the web (US launch, September 2024).
Technologies
- LLMs
- Python
- Cloud infrastructure
Skills
- Conversational AI
- LLM systems
- Production ML
Related projects
What I built
- Built Dawn largely from the ground up as one of its core engineers, from early architecture through production launch.
- Built both of Dawn's surfaces, the myAir app experience and the web experience, on a single shared backend.
What Dawn is
Dawn is ResMed's generative-AI sleep-health assistant. It launched September 30, 2024 for CPAP therapy users in the US, inside the myAir app and on the web. Users ask it questions about their sleep, their therapy, and their equipment in plain language, any hour of the day, and it answers conversationally using their own therapy data and connected health metrics.
ResMed positions Dawn as a wellness concierge, not a medical device: it explains, coaches, and troubleshoots, and it does not diagnose or prescribe. The product draws on ResMed's cloud platform and a respiratory-medicine dataset ResMed publicly describes as 23 billion de-identified nights of data.
What it does for users
Typical sessions cover therapy questions (mask fit, comfort, troubleshooting), interpreting myAir scores and sleep data, and sleep-health coaching with personalized nudges. ResMed has reported strong early adoption publicly: within months of launch roughly a quarter of myAir visitors had started a Dawn session, and queries to human contact centers dropped by about 40 percent. Those are ResMed's product-level numbers, not personal claims.
My role
I built Dawn largely from the ground up as one of its core machine-learning engineers, from early architecture through production launch. One backend powers both surfaces, the myAir app experience and the web experience, and I built both.
The engineering foundation came out of platform work I did as a founding engineer on ResMed's internal GenAI platform: retrieval-augmented generation infrastructure, agent architectures, evaluation pipelines, and the security hardening a healthcare launch requires. Finer-grained contribution detail is being extracted from repository history and will land here once confirmed.
Sources
Product facts above come from ResMed's public materials: the September 2024 product announcement, resmed.com and myAir product pages, and ResMed leadership interviews and earnings commentary from 2024 through 2026.
notes
- Dawn is a ResMed product built by a team. This page describes the product from public sources and my role at a public-safe level, without internal implementation details.