ML Design Examples

• Recommendation system: candidate generation → ranking → re-ranking, collaborative filtering, content-based, embeddings, two-tower model
• Search ranking: query understanding, retrieval (BM25, dense retrieval), learning to rank (pointwise, pairwise, listwise)
• Ads click prediction: feature engineering (user, ad, context), real-time bidding, calibration, explore-exploit
• Fraud detection system: real-time streaming, feature pipelines, imbalanced classification, human-in-the-loop
• Content moderation: multi-modal classification (text + image), policy-as-code, escalation workflows
• Conversational AI system: intent detection, dialogue management, retrieval-augmented generation, guardrails
• Large-scale image search: embedding extraction, approximate nearest neighbour (ANN), indexing, serving