Applied Scientist

Role Mission

As HAI's LLM Post-Training Applied Scientist, you will own the reinforcement learning and on-policy distillation pipeline that transforms raw model capability into reliable, safe clinical behavior. Your post-training methods will directly determine how our AI agents reason through complex clinical scenarios, handle safety-critical decisions, and ultimately impact millions of patient interactions. This role exists because post-training is where capability becomes trustworthiness—and in healthcare, that's everything.

What You Will Accomplish

Own your first major outcome: By day 90, you will have shipped a post-training improvement that meaningfully advances model performance on a critical clinical capability (clinical reasoning, safety alignment, or task completion), evaluated the gains rigorously, and contributed that learning to our post-training roadmap.

Drive lasting impact: At 12 months, you will have designed and shipped multiple post-training methods that measurably improve our models' clinical safety and reasoning, built reusable infrastructure (reward models, verifiers, evaluation frameworks) that accelerate future post-training work, published your research or contributed to HAI's intellectual property, and directly shaped how our deployed models behave in production healthcare environments.

The Team

You'll work alongside ML researchers, engineers, clinicians, and safety experts who are obsessed with building trustworthy AI. This is a highly technical team that values rigor, collaboration across disciplines, and solving the hardest problems in AI safety and alignment. You'll have direct influence on model architecture and training decisions.

What You'll Do

• Design and implement RL and OPD post-training methods including RLHF, RLVR, on-policy distillation, and novel approaches tailored to healthcare AI—selecting the right methods for different clinical reasoning and safety challenges

• Build and evaluate reward models, verifiers, and LLM-as-judge pipelines that provide reliable training signals for post-training, ensuring they capture what truly matters in clinical contexts (accuracy, safety, patient experience)

• Develop conversational AI environments and simulations for healthcare RL training—creating synthetic clinical scenarios and datasets that enable safe, scalable post-training without relying solely on human feedback

• Automate post-training research loops using agents and tooling to systematically explore hyperparameters, methods, and data strategies—turning post-training into a scientific, reproducible process