Decision Intelligence Engineer - Next Best Action

Decision Intelligence Engineer – Next Best Action role focused on designing, training, and evaluating reinforcement learning policies for Humana’s Next Best Action platform in healthcare. The role involves developing decision-making models and implementing RL methods (e.g., PPO, A3C, DQN, CQL, Decision Transformer), framing member decision problems as MDPs/Partially Observable MDPs, and integrating clinical eligibility rules and program objectives into learning objectives. Responsibilities include building simulation/backtesting environments (discrete-event simulation, Monte Carlo), diagnosing and remediating failure modes (policy collapse, credit assignment issues, distributional shift), establishing automated evaluation gates within Databricks workflows, and instrumenting training with MLflow. You will own the nightly Databricks training workflow, perform feature engineering with Databricks/PySpark, manage model artifacts in MLflow, and collaborate with data engineers, rules engines, and platform architects to ensure production-grade, auditable decisioning. Preferred and required backgrounds include multi-agent frameworks, constrained optimization methods (LP/MIP/Lagrangian relaxation), experience in regulated domains (healthcare/insurance), and familiarity with simulation environments and observability tooling. This role is remote (nationwide in the US) with possible travel to Humana offices for training or meetings.