Genius Sports is enabling a new era of sports for fans worldwide through next-gen technology and live data. They are seeking a Staff Applied AI Engineer to lead the architecture of a multi-agent LLM reasoning layer, focusing on converting multimodal evidence into validated outputs and driving performance and cost optimization.
Responsibilities:
- Own the end-to-end technical direction for the multi-agent, multimodal platform that converts broadcast/radio inputs into validated, structured outputs from prototype to production
- Design and evolve the agent architecture (agent boundaries, interfaces, and orchestration patterns), including evidence fusion, traceability/provenance, and schema-first outputs with versioning and backward compatibility
- Define reliability standards for probabilistic systems: confidence scoring and gating, escalation paths for low-confidence segments (including optional human-in-the-loop), and safe correction/overwrite semantics for live outputs
- Drive performance and cost optimization, selecting routing strategies (lightweight vs heavy models), and implementing batching/caching/retries that keep quality stable under real-time constraints
- Partner across product, platform, and domain experts to translate ambiguous sport scenarios into system logic
- Champion continuous improvement by evaluating new technologies, tools, and approaches where they provide clear value
- Mentor and coach engineers across teams, supporting technical growth and raising the overall engineering bar