AI Scientist / Chief Medical AI Scientist — Metabolic Health & Wearable AI

Yuwell Kailite (鱼跃凯立特) | A subsidiary of Yuwell Medical (SZSE: 002223)Location: Nanjing/Shanghai, China (Relocation support & visa sponsorship available)Employment Type: Full-timeReports To: CTO / VP of AI & Digital HealthTeam Size: Building a 30-person AI team from 0 → 1About YuwellYuwell Medical (002223.SZ) is a leading global medical device company headquartered in China, with 2025 revenue of RMB 7.955 billion and a board-level strategy of "Globalization, Digitalization, and Wearables."Our glucose monitoring business (BGM & CGM) is the company's #1 growth engine:CGM revenue more than doubled in 2025Flagship product Anytime 5 CGM: MARD 8.58% (iCGM-level accuracy), 16-day wear, EU MDR IIb certification granted in early 2026Companion app "Anytime": 1M+ downloads, 4.7/5.0 on Apple App Store, #1 in the Medical → Wearable Devices category in ChinaWe are building the medical-grade equivalent of Oura's smart ring — but powered by CGM. The global wearables + AI health market is exploding (Oura: $11B valuation, confidential IPO filing in May 2026). We have what they don't: medical device licenses, tier-1 hospital clinical networks, and a multi-category hardware matrix.The OpportunityWe are hiring a Chief Medical AI Scientist to define and lead the next generation of AI for metabolic health. You will architect foundation models for physiological time-series, fuse multimodal biomarker data (glucose + blood pressure + SpO₂ + heart rate + blood ketone + uric acid), and deploy models that directly impact hundreds of thousands of real users — not a research prototype, but a shipping product.This is a greenfield, zero-legacy-tech-debt role. You will define the technical roadmap from scratch, build a 30-person team, and publish at top venues while pushing models through NMPA / FDA / CE regulatory pathways.Why This Role Is UniqueWhat You Get📊 Data Moat Real-world, large-scale, longitudinal multimodal metabolic data from millions of app users and thousands of CGM sensors — a dataset global AI labs cannot easily replicate in China🔄 Full-Stack Ownership Hardware → App → Cloud, all in-house. Model outputs traceable to user behavior changes; feedback loops drive continuous iteration🏥 Clinical Pathway Active collaborations with top-tier tertiary hospitals and medical schools for clinical validation, NMPA registration, and co-authored publications🚀 Zero Tech Debt No legacy systems. You define the architecture, the stack, and the roadmap🏦 Public Company + Startup Speed Backed by a profitable, dividend-paying listed company, with the agility of a new ventureKey ResponsibilitiesDefine the AI technical vision for metabolic health: time-series forecasting, multimodal fusion, and foundation models for physiological signalsLead core algorithm R&D: glucose prediction (future 2-hour curve), trend alerting, postprandial attribution, Hypoglycemia Risk WarningDrive multimodal fusion: integrate CGM with BP, SpO₂, HR, blood ketone, uric acid using Transformers, LSTMs, state-space modelsCollaborate with clinical KOLs from top hospitals to validate algorithms and design clinical studiesPublish & patent: target top AI/ML venues (NeurIPS, ICML, KDD, MICCAI) and medical journals (Diabetes Care, Nature Medicine, The Lancet Digital Health)Build & mentor the 30-person AI research organization (hiring, technical direction, culture)Ensure regulatory readiness: design for NMPA SaMD, FDA 510(k), CE MDR compliance — safety, interpretability, and robustness by designQualificationsMust-HavePhD in Computer Science, EE, Biomedical Engineering, Statistics, or related fieldStrong publication record at top-tier venues: NeurIPS, ICML, ICLR, KDD, MICCAI, ACL, etc. — especially in time-series analysis, deep learning, or multimodal learningDeep expertise in time-series modeling: LSTMs, GRUs, Transformers, temporal convolutional networks, state-space models (e.g., S4, Mamba)Proficiency in Python, PyTorch/TensorFlow, and modern ML infrastructureAbility to translate ambiguous clinical problems into well-defined AI tasksHighly PreferredExperience with physiological signals: CGM, ECG, PPG, EEG, or similar continuous sensor dataTrack record of taking ML models from research → production → regulatory clearanceFamiliarity with edge AI deployment (on-device inference, quantization, distillation)Experience with federated learning / privacy-preserving ML for healthcareOpen-source contributions or personal research codebaseNice-to-Have (Domain knowledge can be learned on the job)Familiarity with diabetes pathophysiology, CGM clinical guidelines (ADA, CDS), or NMPA AI medical device regulationsPrior work in medical imaging, EHR, or digital biomarkersWe care more about your AI depth than your medical background. Domain expertise will be provided by our clinical advisory team and hospital partners. Many breakthrough medical AI researchers came from non-medical backgrounds (e.g., Vivek Natarajan at Google Health, originally an aerospace engineer).What We OfferCategoryDetailsCompensation Highly competitive compensation package including base salary, performance bonus, and stock options are also alternatives for open discussion Research Budget Full support for R&D with a dedicated research budget covering cloud compute (GPU clusters), data acquisition, conference travel, and open-source contributionsImpact at Scale Your algorithms ship to hundreds of thousands of active users — not a simulation, real patients, real outcomesTeam Building You define the hiring plan and technical culture of a 30-person AI organization Relocation Full package: visa sponsorship, housing allowance, settling-in bonus, flight reimbursement, spouse/child supportClinical Access Direct collaboration with top-tier tertiary hospitals in China for data access, clinical trials, and co-publicationAcademic Freedom Encouraged to publish, open-source, and speak at conferences. We view papers and patents as core IP, not distractions