Principal Machine Learning Engineer, Short-form

Paramount is on a mission to unleash the power of content, and they are seeking a Principal Machine Learning Engineer to lead the technical direction for the Shortform pod. The role involves defining the architecture and modeling strategy for Personalization of Short-form experiences and ensuring long-term user satisfaction through advanced machine learning techniques.

Responsibilities:

• Set Technical Strategy: Own the multi-quarter technical roadmap for short-form ranking, candidate generation, and Post-training RL
• Architect End-to-End Systems: Design the multi-stage ranking architecture spanning retrieval, ranking, re-ranking, and RL-based policy optimization
• Advance reinforcement learning in production. Drive the use of post-training reinforcement learning techniques, including reward modeling, off-policy evaluation, and policy alignmentto improve user satisfaction over long periods
• Cross-Pod Influence: Partner with Content Understanding, ML Platform, Core Science, and Product to align short-form personalization with broader Discovery strategy
• Operate at Scale: Ensure ranking pipelines are high-throughput, reliable, and observable in GCP using TensorFlow/PyTorch
• Mentorship & Talent: Mentor IC1–IC3 engineers, set technical standards across the pod, and grow the next generation of senior ML talent
• Mitigate Systemic Risk: Identify and resolve feedback loops, exposure biases, and filter-bubble dynamics in how short-form content is surfaced

Requirements:

• Minimum: 8+ years of experience in MLE with a track record of setting technical direction for large-scale ranking or recommender systems
• deep expertise in Reinforcement Learning, particularly Post-training RL and long-horizon reward modeling
• proficiency in GCP, TensorFlow, and PyTorch; demonstrated ability to influence technical strategy across multiple teams
• Experience in video-first social or streaming apps
• background in multi-modal signal processing
• published work or recognized contributions in ranking, RL, or recommender systems