Machine Learning Scientist 4 - Generative Models, Evaluation

Track advancements in LLM and generative model research, and translate them into concrete opportunities for the problem space we are working on. Apply state-of-the-art techniques to high-impact problems such as summarization quality and building LLM-based evaluators ("judges") that assess whether generated content meets quality bars. Lead end-to-end ML development on these problems: research, model training, and evaluation. Optimize generated content for member impact, treating content generation and audience targeting as coupled problems — drawing on recommendation systems thinking to ensure generated assets are optimized for the members they're meant to reach. Partner with ML scientists and engineers to integrate models into business applications and platforms. Translate complex technical concepts for both technical and non-technical stakeholders to inform strategic decisions. Act as a subject matter expert, working cross-functionally to identify high-impact opportunities and define and execute roadmaps. Demonstrated habit of tracking the LLM/generative model literature and a knack for spotting which advances are relevant to real business problems. Strong foundation in ML and Deep Learning, especially in Natural Language Processing / Understanding and Generation. Research experience in summarization or generated-text evaluation (e.g., LLM-as-judge methods) is a big plus; background in vision models such as diffusion models is a plus. Expertise in one or more LLM post-training approaches. Experience with recommendation systems, personalization, or other member-facing optimization systems is a strong plus.