Data Scientist – Medical AI

Data Scientist – Medical AI (Computer Vision & Physiological Signals)
About the role
Caare Healthtech Services is building clinical-grade AI screening modules: contactless cardiovascular monitoring from video (rPPG), oral cavity image analysis, and dermatological lesion classification. You'll own models end to end, from data and training through evaluation and deployment-ready inference APIs, in a small team with hard milestone deadlines. Beyond these modules, you'll also contribute to new AI projects as Caare's product portfolio grows.
What you'll do
Train, fine-tune, and benchmark deep learning models for image classification, detection, and segmentation (oral lesions, skin lesions) and for video-based signal extraction (HR, HRV, SpO₂, BP estimation, stress indices)
Build preprocessing pipelines covering image quality gating, ROI detection, motion artifact and lighting correction, and signal denoising
Design rigorous evaluation: sensitivity/specificity, calibration, confidence scoring, and subgroup performance across skin tones, age, and demographics, with honest reporting of limitations
Survey and reproduce recent research, and decide what's viable for production versus what's still a research claim
Package models as low-latency inference services (cloud and edge) with standardised outputs, audit logging, and secure APIs
Write technical documentation covering architecture, training methodology, validation results, and integration
Support clinical validation activities with technical evidence and fast iteration on feedback
Must have
2+ years in applied ML/data science, with models you personally trained and deployed
Strong PyTorch skills, with hands-on work in CNNs/ViTs, transfer learning, fine-tuning foundation models, and data augmentation for small or imbalanced datasets
Solid evaluation discipline: proper train/val/test splits, leakage prevention, statistical testing, calibration, and error analysis
Experience with medical imaging, biosignals, or other high-stakes domains
Ability to read papers critically and translate them into working code
Deployment experience with ONNX/TensorRT or similar, FastAPI, Docker, and at least one cloud platform
Strong plus
rPPG, PPG, or physiological signal processing (filtering, spectral methods, HRV analysis)
Experience with dermatology or oral pathology datasets (ISIC, HAM10000, or similar)
Fairness evaluation across skin types (Fitzpatrick scale)
Familiarity with GDPR, medical device software standards (IEC 62304, EU MDR), or clinical validation protocols
Publications or open-source work in medical AI
Edge/mobile model optimisation (quantisation, pruning)
What we offer
Ownership of core models in a company that retains its technology
Competitive compensation
Flexible/remote working
To apply
Send your CV to contact@caare.in and one example of a model you trained and evaluated, covering what metrics you used and what failure modes you found.

Work arrangement
Hybrid