Design trending livestream discovery
Design a Trending Livestream Discovery System
Design an end-to-end system for a live-commerce platform like Whatnot that identifies and surfaces trending livestreams to users.
Core Assumption
An existing Machine Learning (ML) model for scoring the trendiness or relevance of a livestream is available as a service. Your task is to design the surrounding ML system architecture.
Discussion Points
Your design should cover the following aspects:
- Product Goals & Success Metrics: Define what "trending" means in this context and how the system's success will be measured.
- Data Collection: Identify the events and features necessary to collect for both model input and system operation.
- Candidate Generation: Describe how a pool of potential livestreams will be generated for evaluation.
- Ranking Logic: Explain how business rules, real-time signals, and the pre-existing ML model score will be combined to determine the final ranking.
- System Architecture: Detail the online and offline components of the system.
- Non-Functional Requirements: Address latency, data freshness, and scaling considerations.
- Quality Control: Outline mechanisms to prevent stale, low-quality, spammy, or ended streams from being recommended. Specifically, how would you ensure a stream shown in the trending feed is actually still live?
- Operational Aspects: Discuss strategies for experimentation (A/B testing), monitoring, and failure handling.
AI interview prep
Core dilemma
Balancing real-time data freshness and low-latency serving for dynamic content against the complexity and cost of a scalable, fault-tolerant ML inference pipeline.
3 tradeoffs and 3 stress tests locked
Interview prep
Unlock the interviewer X-Ray rubric, tradeoff matrix, stress tests, clarifying questions, and key concepts.
Design trending livestream discovery
Design a Trending Livestream Discovery System
Design an end-to-end system for a live-commerce platform like Whatnot that identifies and surfaces trending livestreams to users.
Core Assumption
An existing Machine Learning (ML) model for scoring the trendiness or relevance of a livestream is available as a service. Your task is to design the surrounding ML system architecture.
Discussion Points
Your design should cover the following aspects:
- Product Goals & Success Metrics: Define what "trending" means in this context and how the system's success will be measured.
- Data Collection: Identify the events and features necessary to collect for both model input and system operation.
- Candidate Generation: Describe how a pool of potential livestreams will be generated for evaluation.
- Ranking Logic: Explain how business rules, real-time signals, and the pre-existing ML model score will be combined to determine the final ranking.
- System Architecture: Detail the online and offline components of the system.
- Non-Functional Requirements: Address latency, data freshness, and scaling considerations.
- Quality Control: Outline mechanisms to prevent stale, low-quality, spammy, or ended streams from being recommended. Specifically, how would you ensure a stream shown in the trending feed is actually still live?
- Operational Aspects: Discuss strategies for experimentation (A/B testing), monitoring, and failure handling.