Software engineer · Founder, Aevox · Exercise physiologist
I build AI software end to end and work forward-deployed to run it with the customer. I founded Aevox, AI analytics for metabolic testing, deployed in five labs, from a CommonSpirit Health / EXOS performance center to independent labs. I run every deployment myself: the first sales call, each lab’s raw exports as they actually arrive, and the fix that ships the same week. As an exercise physiologist, I can read the test the software analyzes. NYU Computer Science; three years building analytics and trading credit at Barclays.
What I'm Building
Company
AI analytics for metabolic testing. It reads each lab’s raw cart export; when the rules can’t, a model proposes a column map that must pass physics checks on the data. A deterministic engine finds the thresholds, and nothing reaches the athlete until a physiologist confirms. Deployed in five labs; in production at Hybl Performance Center.
How it’s built → aevox.health →Projects
Longitudinal voice monitoring for Parkinson's. Short conversational check-ins establish a patient-specific speech baseline, then detect drift over time — surfaced to clinicians as trend lines, not a snapshot. Self-referenced monitoring, not a diagnostic classifier. Won Healthcare × AI NYC Hackathon 2026.
parsel.health →Predicting metabolic substrate use from consumer wearable signals. Physiology-grounded ML.
GitHub →Research Tools
Auditable grade matrix for every consumer-wearable claim — VO₂max, HRV, sleep, cuffless BP. Computed verdicts, not asserted.
Open tool →135 biomarkers, 519 evidence cells linking labs to mortality, CVD, dementia, frailty. Citation-backed, open source.
Open tool →Experience and Education
Built and owned the desk’s analytics stack (Python, SQL, kdb/q) for PnL attribution, volumes, market share and risk. Made markets and managed risk on a $1B book of front-end IG corporate bonds, with real-time coverage of 20+ buy-side accounts. Tradeweb 2023 Top 5 Performer.
B.A. Computer Science, New York University (2022), after two years as a CS major at the United States Military Academy (Dean’s List every semester, Cadet First Sergeant). M.S. Applied Physiology, University of Colorado Colorado Springs (2025 – 2027, GPA 4.0): mitochondrial bioenergetics and metabolic-cart validation.
Selected Research
A sex-dimorphic thermodynamic-framework analysis of the MoTrPAC rat endurance training dataset. 85 redox-relevant genes, 42,770 observations across 19 tissues, 5 omic layers, 4 training time points. Redox buffering scales significantly sub-linearly with electron transport system expansion in both sexes — with distinct quality-control architectures (mtUPR / mitophagy in males, AMPK in females).
First published independent validation of the MGC Meridian, and the first MGC-vs-Douglas comparison during maximal treadmill running. Nine athletes, 17 paired observations, Bland–Altman across VE, VO₂, VCO₂, RER. Headline: the cart overestimates VO₂ by 6.5% (p<0.001) and underestimates RER by 4.6% (p<0.001) at maximal effort — a non-trivial bias for any lab using MGC for VO₂max-based decisions.
Engineering notes live here, starting with how Aevox reads a metabolic test. I also write about physiology and building on Substack, with shorter thinking on X.
Engineering notes → Substack → X →