Associate Director, Reinforcement Learning; ML

Associate Director, Reinforcement Learning (ML)

Join Amgen Mission of Serving Patients
At Amgen, if you feel like you're part of something bigger, it's because you are. Our shared mission to serve patients living with serious illnesses drives all that we do.
Since 1980, we've helped pioneer the world of biotech in our fight against the world's toughest diseases. With our focus on four therapeutic areas Oncology, Inflammation, General Medicine, and Rare Disease we reach millions of patients each year. As a member of the Amgen team, you'll help make a lasting impact on the lives of patients as we research, manufacture, and deliver innovative medicines to help people live longer, fuller happier lives.
Our award‑winning culture is collaborative, innovative, and science based. If you have a passion for challenges and the opportunities that lay within them, you'll thrive as part of the Amgen team. Join us and transform the lives of patients while transforming your career.

Lets do this. Lets change the world. In this vital role you will lead Amgen's strategy and execution for Reinforcement Learning from Human Feedback (RLHF) and related reinforcement learning approaches across R&D, medical, operations, and commercial use cases. You will design, implement, and scale RLHF systems to solve real‑world problems that ultimately help us serve patients better and faster.
This role requires deep technical expertise in RLHF and modern machine learning, combined with strong leadership capabilities in stakeholder management, cross‑functional collaboration, and organizational influence. You will be expected to translate complex concepts into clear, actionable strategies for senior leaders and guide teams from idea to impact.

• Lead the design and development of RLHF systems including reward modeling, policy optimization, safety and alignment mechanisms, and evaluation frameworks for large language models and other AI systems.
• Drive hands‑on technical execution, particularly for high‑impact projects, reviewing architectures, experimentation plans, and code, and helping the team navigate scientific and engineering trade‑offs.
• Establish best‑practice pipelines for human feedback, partnering closely with internal customer teams to define feedback protocols, annotation quality standards, and governance for RLHF data.
• Define and track success metrics for RLHF systems, balancing offline and online evaluation, A/B tests, safety and robustness criteria, and business or scientific outcomes.
• Collaborate across Amgen leaders to ensure RLHF solutions are aligned with strategy, compliant with policy, and integrated into real workflows.
• Partner with Data, Platform and Technology teams to ensure that RLHF workloads are supported by scalable data platforms, model hosting, experimentation infrastructure, and MLOps best practices.
• Champion responsible and compliant AI, working with Legal, Compliance, and Information Security to implement governance around human feedback, data usage, model behavior, transparency, and risk management in a regulated environment.
• Communicate insights and influence senior stakeholders, creating clear narratives, roadmaps, and recommendations that help executives understand RLHF trade‑offs, risks, and opportunities.

We are all different, yet we all use our unique contributions to serve and the professional we seek will have these qualifications.

• Doctorate degree and 3 years of Computer Science, IT or related field experience
• Masters degree and 5 years of Computer Science, IT or related field experience
• Bachelors degree and 7 years of Computer Science, IT or related field experience
• Associates degree and 12 years of Computer Science, IT or related field experience
• High school diploma / GED and 14 years of Computer Science, IT or related field experience

• Certifications on Reinforcement Learning (AWS AI, Azure AI Engineer, Google Cloud ML, etc.) are a plus.

• Deep, hands‑on expertise in Reinforcement Learning from Human Feedback (RLHF) and/or advanced reinforcement learning, including reward…