Staff Machine Learning Engineer

Storytelling is, and will always be, the heart of what we do at Skydance. Our vision is to build a thriving entertainment ecosystem where stories and worlds are imagined, brought to life and experienced. We create for every age and audience, anchored by the strength of our content and experiences and the depth of our connection with fans. We are creator-first, audience-focused, tech-forward, globally scaled. Here, you’ll work with talented teams that move fast, think boldly and collaborate with purpose to deliver results. We aim high, champion the best ideas and win together, earning the trust of our employees, creators, partners and audiences every day. If you’re energized by the opportunity to help build the next-generation global media and entertainment company powered by creativity and technology, we’d be thrilled to welcome you aboard. LET’S GO! ㅤ Staff Machine Learning Engineer We are looking for a passionate Staff Machine Learning Engineer to build, run and scale the production inference platform behind search and personalization for the global streaming app Max as well as ongoing and future streaming initiatives. Every recommendation, search result and personalized row a viewer sees is served by a model running on this platform, and it has to answer in milliseconds for 100+ million users worldwide. The team, ML Infrastructure & Tools (MLIT), owns the deployment, lifecycle and inference of models for the Search & Personalization organization. We build and operate a shared, self-service inference platform on Kubernetes that serves models across multiple frameworks and runtimes behind one common control plane and model lifecycle. You will be the technical owner of that serving runtime. You are excited about working in a fast[1]moving environment that fosters engineering and operational excellence, and you believe a platform is only done when data scientists can ship models on it without you in the critical path. Roles & Responsibilities Own the architecture and roadmap of the model serving runtime, supporting multiple ML frameworks and inference engines behind one control plane Lead large-scale migrations and rollouts of production models onto the platform with zero viewer-facing regressions Build and maintain serving backends, model configurations and batching strategies that meet strict tail-latency targets at high request volume Design autoscaling, capacity management and multi-region resilience for latency[1]critical serving across Max, Discovery+ and other WBD offerings Drive observability, cost optimization and performance tuning across the serving fleet, including hardware selection, GPU and CPU utilization, and model optimization Define the contract between model training and serving so a validated model artifact deploys to production through standard, automated paths Establish per-model SLOs with model owners and build the telemetry that proves them