Location: New York City, San Francisco, or Austin
This is an applied AI engineering role, sitting between the ML infrastructure team and the product engineering team. The Applied AI Engineer takes model capabilities and builds production-grade AI systems that operate on healthcare data — clinical documentation coding, denial management, claims processing, and revenue cycle automation.
Day-to-day work is building and improving production AI pipelines, designing evaluation systems to measure whether the AI is working correctly, analyzing human-in-the-loop feedback to improve model performance, and translating complex healthcare workflows into structured prediction problems that AI systems can reliably solve.
Primary language: Python — production systems with APIs, structured data, async workflows, testing, logging, and observability.
PyTorch — confirmed as the ML framework.
LLM application stack: agent frameworks, tool calling, retrieval/RAG, structured outputs, prompt design, model evaluation.
Evaluation and benchmarking infrastructure: annotation workflows, experiment tracking, regression testing.
Model training: fine-tuning, supervised fine-tuning, reward modeling, distillation, synthetic data generation (listed as strong signals, not requirements).
Healthcare data: clinical documentation, medical coding (ICD/CPT), claims, denials, payer policy, revenue cycle workflows.
REQUIRED
• Production AI/ML system experience — building and operating AI systems that run in production, not just research notebooks or experiments.
• Strong Python proficiency for production systems — APIs, async workflows, structured data processing, testing, logging, observability. Not just scripting.
• PyTorch proficiency.
• ML fundamentals.
• LLM application experience — at least one of: agent frameworks, tool calling, retrieval/RAG, structured outputs, prompt design, or model evaluation. Depth in any of these is valued.
• Experience building evaluation systems, benchmarks, or annotation workflows for AI systems.
PREFERRED
Healthcare domain experience — clinical documentation, medical coding, claims, denials, payer policy, or revenue cycle management..
Experience framing messy real-world workflows as structured prediction problems — multi-label classification, ranking, extraction, and decisioning tasks.
• Fine-tuning, reinforcement learning, reward modeling, distillation, or synthetic data generation experience.
• Experience with high-stakes, regulated, or auditable AI systems where explainability is required.
• Startup or research-to-production environment experience.
• Experience building systems over large-scale unstructured or semi-structured data.