SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
Abridge is an AI-powered healthcare platform that transforms clinical conversations into structured documentation in real-time. Founded in 2018, the company has built enterprise-grade technology with deep EMR integrations and is expanding into clinical workflows including coding, clinical decision support, and care navigation.
You will serve as Product Lead for Abridge's foundational models and post-training initiatives, partnering closely with ML Science to translate proprietary clinical data into durable model capabilities. This role sits at the intersection of model science, platform strategy, and clinical product delivery—not a traditional feature-PM role, nor a research-program-manager role.
Key responsibilities include: setting product strategy for Abridge's model family with explicit tradeoffs across capability, quality, latency, cost, safety, and controllability; owning the path from research to production impact by defining hypotheses, milestones, decision gates, and success metrics; turning proprietary de-identified conversations, clinician edits, EHR context, and care actions into training and feedback signals; building frameworks for when to use frontier models, open models, prompted workflows, or Abridge-trained models; partnering with the Evals team on promotion criteria and product-relevant capabilities; creating tight learning loops with product teams to convert production failures and clinician feedback into training priorities; driving cross-functional execution across ML Science, ML Engineering, Product Engineering, Clinical Science, Data, and product pods; and communicating complex research choices to executives and product teams.
You will need 7+ years of product management or closely related experience with substantial ownership of ML-powered products, model platforms, or AI infrastructure. You must have a track record of turning ambiguous technical capabilities into shipped products with measurable outcomes, strong working knowledge of the modern model-development lifecycle (data strategy, fine-tuning, preference optimization, evaluation, inference, experimentation, production monitoring), and the technical judgment to reason with ML scientists and engineers about training objectives, reward design, data quality, model selection, scaling, latency, and serving cost. Strong product judgment about proprietary versus external models, experience creating clarity across multiple teams with defined decision rights and sequencing, a high bar for evidence and safety (especially critical in clinical AI), and excellent written and verbal communication are essential. Bonus experience includes direct work on LLM post-training, reinforcement learning, and preference optimization.
