About the job
At Genius Sports, we’re revolutionizing the sports experience for fans around the globe by leveraging cutting-edge technology and unparalleled live data, delivering immersive, interactive, and personalized experiences like never before. Discover more at geniussports.com.
About the Role
We are seeking a skilled Senior Applied AI Engineer to develop production-grade, multimodal systems (audio/video/text) that transform broadcast and radio feeds into structured, real-time signals and event candidates. You will play a pivotal role in implementing and enhancing “agentic” components, such as sensor agents and decision logic, which power innovative products like Audio Intelligence, semi-automated broadcast-to-data tagging, and highlight/momentum signals.
Your technical expertise and pragmatic problem-solving approach will be essential as you collaborate within a team that values Agile delivery principles and continuous improvement. You will adopt a data-driven mindset, embracing the principles of continuous experimentation and validation.
Key Responsibilities
- Develop and maintain multimodal agents:
- Audio sensor agents (acoustic events, sentiment analysis, alignment)
- Visual sensor agents (scorebug/overlay interpretation, basic visual cues)
- Specialist and decision logic components (structured event outputs, confidence scoring, traceability)
- Implement streaming-friendly pipelines encompassing chunking, normalization, time synchronization, asynchronous execution, and robust retry/backoff strategies for model/tool calls.
- Develop prompt-as-code with strict JSON contracts, schema validation, and deterministic post-processing to enhance system stability.
- Enhance system resilience against noisy inputs by:
- Designing fallback behaviors for degraded modes
- Incorporating guardrails and confidence thresholds
- Instrumenting traces and metrics for latency, cost, and accuracy
- Collaborate with product, platform, and domain leads to translate sports rules and edge cases into validation logic and integrate outputs into downstream consumers (tagging, live feeds, analytics).
- Contribute to the evaluation workflow by adding test cases, failure mode categories, and regression checks for prompts and model routing.
- Stay updated with emerging generative AI technologies, tools, and trends.
