AI Engineer
Core
Build real-time AI systems that transform live sports broadcasts (video, audio, commentary) into structured game context and auto-generated insights for products like automated highlights and augmented broadcasts.
Role type
Senior IC software engineer (real-time streaming & multimodal AI)
Builds
Real-time pipelines processing noisy, asynchronous inputs under tight latency constraints to produce reliable outputs in production.
Domain
Sports media, streaming, distributed systems, multimodal AI
Deliverable
production ML models
Required skills
System design, testing, observability, performance engineering, building streaming/distributed systems, handling uncertainty in model outputs, designing evaluation frameworks, end-to-end system ownership, architectural decision-making
Preferred skills
Experience with multimodal AI models, working through ambiguity, building systems from first principles
Technologies
(Not explicitly listed)
Responsibilities
Design, build, and operate real-time streaming systems that convert live sports broadcasts into structured events, attributes, and insights; Ingest and synchronize signals from video, audio, commentary, and data feeds under real-world latency and reliability constraints; Integrate modern AI models into production systems, including multi-step pipelines where model outputs are composed, validated, and refined; Design and implement evaluation and observability frameworks to measure system quality, including offline benchmarks, real-time metrics, and regression testing; Develop mechanisms for identifying and handling uncertainty in model outputs, such as confidence scoring, validation checks, and human-in-the-loop workflows; Own systems end-to-end from early prototypes through production, making key architectural decisions around scalability, reliability, and performance; Collaborate closely with product and data partners to shape, prioritize, and ship high-impact systems; Mentor engineers and help establish best practices for building and operating applied AI systems in production
