Senior Applied AI Engineer

The opportunity

Build and deploy the intelligence behind Orion and Risk Radar, with support for Eterna where prioritized. Combine rigorous modeling, strong software engineering and practical AI workflows to solve complex insurance problems. We need someone who can take a business question through evaluation to a reliable production capability, not just create a compelling demonstration.

What you'll do

• Develop actionable risk models. Build provider-behavior, anomaly-detection and risk-scoring approaches using claims and related authorized data. Engineer features and explore entity resolution and graph methods where useful.
• Validate rigorously. Address limited labels, imbalance, temporal and entity leakage, calibration and false positives. Evaluate usefulness against investigation outcomes rather than a single aggregate metric.
• Engineer for production. Build tested inference and feature pipelines, integrate model outputs with applications and optimize reliability, latency and processing cost.
• Own model operations. Implement reproducibility, versioning, monitoring, drift detection and controlled release and rollback practices with engineering partners.
• Build bounded AI workflows. Develop evidence-gathering, extraction and review-support agents with traceable sources, permissions, evaluation, human approval where required and safe failure handling.
• Work with domain specialists. Partner with clinical, insurance and customer-delivery colleagues to interpret signals, investigate errors and improve explainability and practical usefulness.
• Protect data and decision quality. Document assumptions, limitations and experiments. Use only authorized data and approved training processes; model outputs support human investigation and do not independently adjudicate claims.

What you bring

• 5+ years of relevant applied AI, machine learning or ML engineering experience, with evidence of deploying and maintaining production models.
• Strong Python and SQL; practical experience with relevant ML frameworks, feature engineering, statistical evaluation and cloud-based data processing.
• Solid software engineering habits, including testing, version control, code review, deployment and observability.
• Ability to identify data leakage, choose meaningful baselines and make clear trade-offs among signal quality, interpretability, operational reliability and cost.
• Independent problem-solving and clear communication with both technical and domain colleagues; able to own an implementation from experimentation through release.

Also valued

• Insurance, healthcare, fraud, risk analytics or payment-integrity experience.
• Graph analytics, entity resolution, weak supervision, limited-label learning or interpretable risk modeling.
• Evaluated LLM, retrieval or agent workflows and familiarity with containerization, orchestration and model-lifecycle tooling.

Your first 90 days

1. First 30 daysUnderstand the data, reproduce a core pipeline and document its current evaluation and operational behavior.
2. By 60 daysEstablish stronger validation or monitoring and test a focused model or workflow improvement against agreed baselines.
3. By 90 daysAdvance a scoped improvement through controlled validation and, where release approvals permit, production deployment with monitoring and documentation.

How success is measured

Actionable-signal quality, false-positive reduction, reproducibility, stability, inference cost and dependable delivery of approved capabilities.

How you work with the team

AI leadership owns the roadmap and modeling standards. Partner with Engineering on integration and infrastructure, and with clinical and delivery teams on validation. This role is not a substitute for day-to-day engineering leadership.

What to send

Include a resume and examples of production AI work, with your contribution clearly stated. Be prepared to discuss validation, leakage, explainability, deployment decisions and operational lessons from a real system.