Playbook / AI system design
AI system design
AI/ML platform design — hellointerview depth
| # | Title | |
|---|---|---|
| 01 | Design an LLM inference serving platform at scale | Study → |
| 02 | Design a retrieval-augmented generation (RAG) platform at scale | Study → |
| 03 | Design an agent orchestration platform with real tool use | Study → |
| 04 | Design a feature store / fine-tuning data pipeline | Study → |
| 05 | Design a content moderation and safety system for generated content | Study → |
| 06 | Design a multimodal search / recommendation system | Study → |
| 07 | Design an LLM evaluation and observability platform | Study → |
| 08 | Design a fine-tuning / RLHF training pipeline at scale | Study → |
| 09 | Design a multi-tenant AI platform | Study → |
| 10 | Design a sandboxing architecture for AI agent code execution | Study → |
| 11 | Design an on-device / edge AI inference architecture | Study → |
| 12 | Design a training-data provenance and IP-risk architecture | Study → |
| 13 | Design durable execution for long-running AI agents | Study → |
| 14 | Design a ChatGPT-style conversational service | Study → |
| 15 | Design an AI coding assistant (Copilot-style) | Study → |
| 16 | Design an LLM-powered customer support assistant | Study → |
| 17 | Design LLM application security against prompt injection | Study → |
| 18 | Design an AI data flywheel and human-feedback platform | Study → |
| 19 | Design a model release, canary, and rollback platform | Study → |
| 20 | Design persistent AI memory and personalization | Study → |
| 21 | Design a real-time voice AI assistant | Study → |
| 22 | Design enterprise PDF Q&A with page citations and grounding | Study → |
| 23 | Design enterprise hybrid retrieval with access-aware ranking | Study → |
| 24 | Design identity and access management for AI agents and MCP tools | Study → |
| 25 | Design a real-time fraud and risk decisioning system | Study → |
| 26 | Design a multi-agent collaboration evaluation scorecard | Study → |
| 27 | Design a real-time wait-time prediction service with an agent and interactive provenance | Study → |
| 28 | Design an enterprise LLM gateway (unified multi-provider platform) | Study → |
| 29 | Design a distributed GPU training job scheduler with preemption and checkpointing | Study → |
| 30 | Design a unified query engine across dispersed data sources (email, calendar, docs, chat) | Study → |
30 of 30 · total playbook 138