David Clift-Reaves
Chief Technology Officer at Happy Health
Austin, Texas, United States
About David Clift-Reaves
David Clift-Reaves is Chief Technology Officer at Happy Health. Location: Austin, Texas, United States.
Light through a finger. Current across skin. Heat leaving the body. I turn physics inside tissue into medical evidence — two FDA 510(k) clearances on our smart ring, and I own the whole chain. Getting a trustworthy number out of any of that is a biophysics problem before it's an engineering problem, and getting a regulator to accept it is a third problem again. Four links, and I've run all of them at once: → Biophysics. Optical emitter/detector geometry and wavelength selection, so the light path actually samples perfused tissue. Electrodermal and bio-impedance sensing. Skin-versus-ambient thermometry. Motion treated as physiology, not noise. Optical performance reported across all skin tones, not hidden in an aggregate. → Embedded. I set the device-to-cloud architecture in 2019, before there was a product: a stream format carrying what's needed to turn raw samples into calibrated units, integrity-checked bulk transfer over BLE, and storage sized to keep full-rate signal rather than discard it. Microamp power budgets. Signed firmware. A trustworthy clock — a physiological record with an unreliable timebase isn't evidence. → Physiological pipeline. Every channel calibrated, with calibration carried inside the recording. Traceable references per ISO 80601. All sources normalized to a common rate, artifact-rejected, quality-gated, and paired with clinical gold standards — polysomnography, arterial line, reference oximetry — under design controls. → AI. I lead the algorithms team and still ship code: sleep staging, SpO2, respiratory rate, activity, stress. I wrote regulatory accuracy gates into CI and brought algorithms under design control — what lets an AI-bearing medical device keep evolving without breaking its clearance. Why it compounds: we keep raw signal, not processed metrics — the difference between a JPEG and a RAW file. That is what makes it data physicians, and smart AI, can actually learn from. New biomarkers can be pursued against data already in hand, with no new hardware and no re-run studies. An AI roadmap here is a modeling problem, not a data-collection problem. Earlier: optical hydration sensing at LVL, and cofounding Supermechanical, where Twine helped start the consumer IoT revolution.
Skills
Additional experience
Cofounder & VP of Engineering
Happy Health
October 2019 – December 2020
Education
- Bachelor's Degree, The University of Texas at Austin
- Astronomy — research affiliation, The University of Texas at Austin