Lead Data Scientist

Responsibilities:

• Own the data problem formulation for AI systems, defining how historical decision data, claim outcomes, and domain signals are structured, labelled, and mined to drive model accuracy and continuous improvement.
• Lead the design of training data strategies: extracting signal from complex healthcare datasets, building high-quality labelled datasets from real-world coding decisions and denial outcomes, and prioritising annotation effort for maximum model impact.
• Drive high-cardinality data mining across large code and payer spaces, identifying rare but high-value patterns that improve autonomous coding performance.
• Own the architecture and evolution of core AI platform spanning real-time inference, batch processing, and model training pipelines across traditional ML and Gen AI workloads.
• Architect multi-modal AI systems combining unstructured clinical text, structured EHR data, and domain knowledge for comprehensive healthcare coding automation.
• Lead technical evaluation of AI tools, platforms, and vendors; own build vs. buy decisions for key AI infrastructure components.
• Mentor and develop AI engineers across NLP, traditional ML, and Gen AI specialisations; lead architecture reviews and establish engineering best practices.
• Partner with Data Engineering, Gen AI NLP, and Product teams to align data and model training strategy with product requirements.

Requirements:

• 8+ years building and deploying production AI/ML systems with demonstrated emphasis on solving the underlying data problem, not just model selection or infrastructure.
• Deep experience working with complex, domain-specific data at scale: extracting signal from noisy real-world datasets, building training data from historical decision records, and handling high-cardinality label spaces.
• Deep expertise across multiple AI domains (NLP, traditional ML, Gen AI) with production deployments in each.
• Expert-level proficiency in PyTorch or TensorFlow; experience with model training at scale.
• Proven track record on AI/ML projects where data quality, labelling strategy, and training data curation were the primary challenge, not just model architecture or deployment.
• Experience with cloud-native AI architectures on AWS, Azure, or GCP; containerised deployments (Docker, Kubernetes) for AI workloads.

Preferred:

• Experience building self-improving AI systems: active learning pipelines, feedback loop architectures that connect outcome signals back to model updates, or reinforcement learning from human feedback (RLHF) / outcome-based reward signals.
• Data mining experience at scale across high-cardinality feature spaces, finding rare but high-value patterns across large taxonomy or code spaces.
• Hands-on experience with domain-specific AI systems that learn from historical decisions: e. g., autonomous decision engines, recommendation systems, or classification pipelines that adapt from labelled outcomes over time.
• Experience with MLOps tooling: CI/CD for ML, model versioning, monitoring, and deployment governance.
• Experience in healthcare technology, medical AI, or regulated industries including HIPAA-compliant system design and audit trail requirements.
• Familiarity with healthcare data standards (HL7 FHIR) or medical coding systems (ICD, CPT).
• Experience with continual learning or domain adaptation for specialised vertical AI applications.
• Knowledge of FDA guidelines for AI/ML as a medical device (SaMD) or equivalent regulated AI frameworks.
• Experience building interpretable AI systems with explainability requirements for clinical or compliance stakeholders.
• Experience with ensemble models, meta-learning, or multi-modal architectures combining structured and unstructured data.