{
  "generated_at": "2026-09-11",
  "description": "Curated high-quality AI Agent, LLM, and RAG interview-experience/source candidates collected from public web sources. Entries are summaries with source links or source lookup hints, not copied source content.",
  "criteria": {
    "minimum_score": 4,
    "license_note": "Use entries as citations and summarized leads. Do not copy long platform content into repository documents.",
    "source_types": {
      "first_hand": "带明确面试过程或个人项目追问的一手复盘",
      "aggregation": "多篇面经或多轮记录的二次整理，只用于发现趋势",
      "reference": "题库、指南或工程资料，不标记为真实面经"
    }
  },
  "items": [
    {
      "id": "github-keepkeen-agent-engineering-notes",
      "platform": "GitHub",
      "title": "KeepKeen Agent 面经、源码分析与论文证据库",
      "company": null,
      "published_at": "2026-09-08",
      "source_url": "https://github.com/keepkeen/keepkeen.github.io",
      "score": 5,
      "source_type": "reference",
      "confidence": "high",
      "verified_at": "2026-09-11",
      "topics": [
        "Evidence Ledger",
        "Role Taxonomy",
        "Coding Agent Source Analysis",
        "Agent Benchmarks",
        "Failure Taxonomy",
        "Agentic RL"
      ],
      "summary": "持续更新的研究型资料，突出证据等级、版本边界、Coding Agent 源码对比、Agent 评测基准、多 Agent 失败分类和 Agentic RL。"
    },
    {
      "id": "github-aigc-interview-book",
      "platform": "GitHub",
      "title": "AIGC-Interview-Book",
      "company": null,
      "published_at": "2026-09-08",
      "source_url": "https://github.com/WeThinkIn/AIGC-Interview-Book",
      "score": 5,
      "source_type": "reference",
      "confidence": "high",
      "verified_at": "2026-09-11",
      "topics": [
        "AgentOS",
        "AgentOps",
        "Harness Evaluation",
        "Enterprise Agent Platform",
        "Self-evolving Agent",
        "Agentic RL"
      ],
      "summary": "大型 AIGC 面试知识库，其中 AgentOS、AgentOps、Harness 分层评测、企业平台和自进化 Agent 可作为进阶选题线索；直接复用需遵守 GPL-3.0。"
    },
    {
      "id": "nowcoder-kuaishou-agent-fullstack-second-round-2026-09",
      "platform": "牛客",
      "title": "快手 AI 全栈开发二面凉经（Agent 应用方向）",
      "company": "快手",
      "published_at": "2026-09-07",
      "source_url": "https://www.nowcoder.com/discuss/926274020192841728",
      "score": 4,
      "source_type": "first_hand",
      "confidence": "medium",
      "verified_at": "2026-09-11",
      "topics": [
        "Agent Application",
        "Project Deep Dive",
        "Skill",
        "Tool Use",
        "Full-stack Engineering"
      ],
      "summary": "近期二面题源，面试整体偏 Agent 应用并持续深挖项目细节，适合补充 Skill、工具使用和端到端交付能力。"
    },
    {
      "id": "nowcoder-bytedance-agent-first-round-2026-09-03",
      "platform": "牛客",
      "title": "字节跳动 9.3 Agent 开发一面面经",
      "company": "字节跳动",
      "published_at": "2026-09-04",
      "source_url": "https://www.nowcoder.com/discuss/925342611194286080",
      "score": 5,
      "source_type": "first_hand",
      "confidence": "high",
      "verified_at": "2026-09-11",
      "topics": [
        "Agent Harness",
        "Long-running Task Latency",
        "Tool Reliability",
        "Context Compression",
        "Structured Output",
        "RAG",
        "Algorithms"
      ],
      "summary": "一手一面记录，重点追问 Harness、长任务延迟异常排查、工具参数可靠性、上下文压缩信息损失和结构化评分输出。"
    },
    {
      "id": "nowcoder-kuaishou-agent-first-round-2026-09-03",
      "platform": "牛客",
      "title": "快手 AI 应用开发 / Agent 开发一面",
      "company": "快手",
      "published_at": "2026-09-03",
      "source_url": "https://www.nowcoder.com/discuss/925163549431742464",
      "score": 5,
      "source_type": "first_hand",
      "confidence": "high",
      "verified_at": "2026-09-11",
      "topics": [
        "AI Coding Quality",
        "Tool Result Cache",
        "Long-term Preference",
        "Data Privacy",
        "Concurrency",
        "Thread Pool"
      ],
      "summary": "一手一面记录，全程围绕实习项目与 AI 使用习惯，新增 AI 代码质量兜底、工具结果缓存、长期偏好存储和敏感信息保护等问题。"
    },
    {
      "id": "nowcoder-alibaba-agent-roundup-2026-08-31",
      "platform": "牛客",
      "title": "阿里 Agent 开发近期面经汇总",
      "company": "阿里巴巴",
      "published_at": "2026-08-31",
      "source_url": "https://www.nowcoder.com/discuss/923740641878646784",
      "score": 4,
      "source_type": "aggregation",
      "confidence": "medium",
      "verified_at": "2026-09-11",
      "topics": [
        "MCP Evaluation",
        "Agent Trace",
        "LangGraph State",
        "Checkpoint Recovery",
        "Skill Progressive Disclosure",
        "KV Cache",
        "Continuous Batching"
      ],
      "summary": "近期汇总型题源，重点补充 MCP Server 评测、Agent 链路追踪、超长任务续跑、评测集构建和 Agent Infra/推理优化联动。"
    },
    {
      "id": "nowcoder-bytedance-agent-roundup-2026-08-28",
      "platform": "牛客",
      "title": "字节跳动 Agent 开发近期面经汇总（含 7-8 月多轮）",
      "company": "字节跳动",
      "published_at": "2026-08-28",
      "source_url": "https://www.nowcoder.com/discuss/922659050167226368",
      "score": 4,
      "source_type": "aggregation",
      "confidence": "medium",
      "verified_at": "2026-09-11",
      "topics": [
        "Agent Harness",
        "Orchestrator",
        "Path Oscillation",
        "Long Context",
        "Tool Retrieval",
        "Coding Agent",
        "Concurrency",
        "Agent Evaluation"
      ],
      "summary": "近期汇总型题源，新增 Harness/Orchestrator 边界、路径震荡、大规模工具路由、Coding Agent 并发冲突和生产评测等工程化考点。"
    },
    {
      "id": "nowcoder-bytedance-agent-first-round-2026-08-12",
      "platform": "牛客",
      "title": "字节 Agent 开发一面（8.12）",
      "company": "字节跳动",
      "published_at": "2026-08-24",
      "source_url": "https://www.nowcoder.com/discuss/921444774287003648",
      "score": 5,
      "source_type": "first_hand",
      "confidence": "high",
      "verified_at": "2026-09-11",
      "topics": [
        "RAG",
        "AI Coding Quality",
        "SSE",
        "WebSocket",
        "Database Index",
        "High Concurrency",
        "Thread Safety"
      ],
      "summary": "一手一面记录，除 RAG 项目外还考察 AI Coding 质量控制、SSE/WebSocket、高并发热点榜单设计、数据库索引和线程安全。"
    },
    {
      "id": "nowcoder-kuaishou-data-agent-first-round-2026-06-16",
      "platform": "牛客",
      "title": "快手 Data Agent 开发一面（06.16）",
      "company": "快手",
      "published_at": "2026-06-16",
      "source_url": "https://www.nowcoder.com/discuss/904419777614049280",
      "score": 5,
      "source_type": "first_hand",
      "confidence": "high",
      "verified_at": "2026-09-11",
      "topics": [
        "LLM vs Agent",
        "ReAct",
        "Agent Framework Design",
        "Data Agent",
        "Project Deep Dive"
      ],
      "summary": "一手 Data Agent 开发面经，从 LLM 与 Agent 边界、ReAct 执行流程一路追问到 Agent 框架模块设计。"
    },
    {
      "id": "github-kaomian-question-bank",
      "platform": "GitHub",
      "title": "kaomian 高频面试题与手撕题库",
      "company": null,
      "published_at": "2026-04-19",
      "source_url": "https://github.com/smile-struggler/kaomian",
      "score": 4,
      "source_type": "reference",
      "confidence": "medium",
      "verified_at": "2026-09-11",
      "topics": [
        "Question Frequency",
        "Top 100",
        "Project Deep Dive",
        "LeetCode",
        "Model Coding"
      ],
      "summary": "以公开面试问题为基础的频次化题库，适合参考 Top100、项目连续追问和 AI 模型手撕题的组织方式；仓库未见明确许可证，不复制原文。"
    },
    {
      "id": "github-agent-engineering-handbook-2026",
      "platform": "GitHub",
      "title": "Agent Engineering Handbook · 2026",
      "company": null,
      "published_at": null,
      "source_url": "https://github.com/EmberRavager/agent-interview",
      "score": 4,
      "source_type": "reference",
      "confidence": "medium",
      "verified_at": "2026-09-11",
      "topics": [
        "Agent Loop",
        "Context Engineering",
        "MCP",
        "Skills",
        "Durable Execution",
        "Harness",
        "Sandbox",
        "HITL",
        "Tracing",
        "Evals"
      ],
      "summary": "面向 2026 Agent 工程岗位的系统复习资料，覆盖 Runtime、可恢复执行、Harness、安全、Trace 和生产级评测。"
    },
    {
      "id": "xhs-taotian-agent-2026-06-29",
      "platform": "小红书",
      "title": "淘天AI Agent一面 问麻了",
      "company": "阿里/淘天",
      "published_at": "2026-06-29",
      "source_url": null,
      "score": 5,
      "topics": [
        "RAG",
        "BM25",
        "OCR",
        "Multi-Agent State",
        "Checkpoint",
        "MCP",
        "LangGraph",
        "SSE"
      ],
      "summary": "最新淘天 AI Agent 一面，集中追问混合检索、扫描 PDF/表格处理、并行 Agent 状态竞争、LangGraph 选型与流式工程。",
      "source_note_id": "6a41d72b000000002100a51e",
      "source_lookup": "小红书站内搜索原标题：淘天AI Agent一面 问麻了"
    },
    {
      "id": "xhs-tencent-ai-app-2026-06-29",
      "platform": "小红书",
      "title": "腾讯ai应用一面面经",
      "company": "腾讯",
      "published_at": "2026-06-29",
      "source_url": null,
      "score": 5,
      "topics": [
        "Multi-Agent",
        "AgentContext",
        "Run Trace",
        "Skills",
        "Context Injection",
        "Risk Control Agent"
      ],
      "summary": "最新腾讯 AI 应用一面，适合补充多 Agent 协作、上下文注入、运行链路追踪和风控 Agent 项目深挖题。",
      "source_note_id": "6a4216040000000016027829",
      "source_lookup": "小红书站内搜索原标题：腾讯ai应用一面面经"
    },
    {
      "id": "xhs-meituan-agent-2026-06-18",
      "platform": "小红书",
      "title": "美团 AI Agent开发 一面面经",
      "company": "美团",
      "published_at": "2026-06-18",
      "source_url": null,
      "score": 5,
      "topics": [
        "PDF RAG",
        "RAGAS",
        "LangGraph",
        "MCP",
        "ReAct",
        "Rerank",
        "JSON Tool Calling",
        "Streaming"
      ],
      "summary": "近期美团 Agent 开发面经，覆盖工业 PDF 跨页表格切片、RAGAS 幻觉评估、心理咨询 Agent 安全流转和工具 JSON 异常兜底。",
      "source_note_id": "6a33f73b00000000220157f1",
      "source_lookup": "小红书站内搜索原标题：美团 AI Agent开发 一面面经"
    },
    {
      "id": "xhs-kuaishou-agent-intern-2026-06-12",
      "platform": "小红书",
      "title": "快手AI Agent开发实习生面经2026.6.12",
      "company": "快手",
      "published_at": "2026-06-12",
      "source_url": null,
      "score": 5,
      "topics": [
        "Agent Basics",
        "ReAct",
        "Multi-model API",
        "Single-Agent",
        "Multi-Agent",
        "RAG Optimization",
        "Coding"
      ],
      "summary": "快手 AI Agent 实习一面，适合作为 Agent 基础、ReAct、统一多模型 API、单/多 Agent 选型和 RAG 优化题源。",
      "source_note_id": "6a2bdd4e0000000016026052",
      "source_lookup": "小红书站内搜索原标题：快手AI Agent开发实习生面经2026.6.12"
    },
    {
      "id": "xhs-meituan-ai-agent-2026-06-04",
      "platform": "小红书",
      "title": "美团AI-Agent工程师面经，看看难度",
      "company": "美团",
      "published_at": "2026-06-04",
      "source_url": null,
      "score": 5,
      "topics": [
        "AI Agent",
        "RAG",
        "Project Deep Dive"
      ],
      "summary": "较新的美团 AI-Agent 工程师面经，适合并入美团公司面经候选池。",
      "source_note_id": "6a2197c3000000002101a9e7",
      "source_lookup": "小红书站内搜索原标题：美团AI-Agent工程师面经，看看难度"
    },
    {
      "id": "xhs-tencent-agent-second-round-2026-05-31",
      "platform": "小红书",
      "title": "腾讯 Agent二面凉经带答案",
      "company": "腾讯",
      "published_at": "2026-05-31",
      "source_url": null,
      "score": 5,
      "topics": [
        "Agent Second Round",
        "Answer Hints",
        "Engineering Deep Dive"
      ],
      "summary": "腾讯 Agent 二面候选，带答案要点，适合补充腾讯二面深挖题。",
      "source_note_id": "6a1c1e5c000000003501dd46",
      "source_lookup": "小红书站内搜索原标题：腾讯 Agent二面凉经带答案"
    },
    {
      "id": "xhs-bytedance-agent-second-round-2026-05-30",
      "platform": "小红书",
      "title": "字节跳动Agent开发岗二面（贼难）",
      "company": "字节跳动",
      "published_at": "2026-05-30",
      "source_url": null,
      "score": 5,
      "topics": [
        "Multi-Agent",
        "LangGraph",
        "Skills",
        "Context Engineering",
        "Agent Evaluation",
        "SFT",
        "Inference Optimization"
      ],
      "summary": "高互动字节 Agent 二面，问题密集，适合补充生产级 Agent 架构、bad case 定位、Skills 体系和模型优化选型。",
      "source_note_id": "6a1ab93a0000000035029e6f",
      "source_lookup": "小红书站内搜索原标题：字节跳动Agent开发岗二面（贼难）"
    },
    {
      "id": "xhs-bytedance-agent-1to3-2026-05-18",
      "platform": "小红书",
      "title": "字节 agent开发 1-3面面经 5月",
      "company": "字节跳动",
      "published_at": "2026-05-18",
      "source_url": null,
      "score": 5,
      "topics": [
        "Doubao",
        "Long Context",
        "A2A",
        "Memory",
        "RAG",
        "Hallucination Mitigation"
      ],
      "summary": "字节 Agent 社招多轮面经，聚焦豆包业务场景、长会话、多 Agent 循环防控和子 Agent 幻觉治理。",
      "source_note_id": "6a0ae15e000000003601db35",
      "source_lookup": "小红书站内搜索原标题：字节 agent开发 1-3面面经 5月"
    },
    {
      "id": "xhs-tencent-ai-app-2026-04-11",
      "platform": "小红书",
      "title": "腾讯 ai 应用开发 一面",
      "company": "腾讯",
      "published_at": "2026-04-11",
      "source_url": null,
      "score": 5,
      "topics": [
        "AI Application Development",
        "Agent",
        "Intern Interview"
      ],
      "summary": "高赞腾讯 AI 应用开发一面候选，适合补充腾讯公司面经。",
      "source_note_id": "69d9d3c8000000001a021c89",
      "source_lookup": "小红书站内搜索原标题：腾讯 ai 应用开发 一面"
    },
    {
      "id": "xhs-xiaohongshu-agent-2026-04-05",
      "platform": "小红书",
      "title": "小红书 AI Agent开发一面",
      "company": "小红书",
      "published_at": "2026-04-05",
      "source_url": null,
      "score": 5,
      "topics": [
        "E-commerce Agent",
        "Master-Worker Architecture",
        "Tool Calling",
        "RAG Indexing",
        "Prompt Debugging",
        "Skills",
        "Latency"
      ],
      "summary": "小红书平台业务相关 Agent 面经，适合补充电商 Agent、主从架构、商品知识库索引和延迟分析。",
      "source_note_id": "69d214f10000000023004de5",
      "source_lookup": "小红书站内搜索原标题：小红书 AI Agent开发一面"
    },
    {
      "id": "xhs-baidu-llm-app-2026-03-28",
      "platform": "小红书",
      "title": "百度 大模型应用开发 一面面经",
      "company": "百度",
      "published_at": "2026-03-28",
      "source_url": null,
      "score": 5,
      "topics": [
        "Multimodal RAG",
        "Table Chunking",
        "Memory Summary",
        "Function Calling",
        "Redis Semantic Cache",
        "Session"
      ],
      "summary": "百度大模型应用开发一面，RAG 工程问题具体，适合补充长表格跨 chunk、视觉 embedding、记忆总结和语义缓存。",
      "source_note_id": "69c7613e000000002202790d",
      "source_lookup": "小红书站内搜索原标题：百度 大模型应用开发 一面面经"
    },
    {
      "id": "xhs-kuaishou-agent-2026-03-15",
      "platform": "小红书",
      "title": "快手AI Agent开发一面",
      "company": "快手",
      "published_at": "2026-03-15",
      "source_url": null,
      "score": 5,
      "topics": [
        "Parent-child Index",
        "BM25",
        "Rerank",
        "Memory",
        "Function Calling",
        "Prompt Injection Defense",
        "RAG Evaluation"
      ],
      "summary": "高互动快手 Agent 面经，RAG + Agent 安全链路完整，适合补充父子索引、BM25、Rerank 和工具调用安全控制。",
      "source_note_id": "69b65422000000001a0312bc",
      "source_lookup": "小红书站内搜索原标题：快手AI Agent开发一面"
    },
    {
      "id": "xhs-bytedance-ai-dev-2026-03-01",
      "platform": "小红书",
      "title": "字节跳动AI开发 一面",
      "company": "字节跳动",
      "published_at": "2026-03-01",
      "source_url": null,
      "score": 5,
      "topics": [
        "AI Development",
        "Agent",
        "First Round"
      ],
      "summary": "字节 AI 开发一面候选，适合补充字节公司面经索引。",
      "source_note_id": "69a43830000000001600a6a1",
      "source_lookup": "小红书站内搜索原标题：字节跳动AI开发 一面"
    },
    {
      "id": "nowcoder-ant-agent-2026-05-20",
      "platform": "牛客",
      "title": "5月20日，蚂蚁智能体与大模型应用 一面",
      "company": "蚂蚁",
      "published_at": "2026-05-20",
      "source_url": "https://www.nowcoder.com/discuss/888874988554448896",
      "score": 5,
      "topics": [
        "Hallucination",
        "Fine-tuning",
        "Skills",
        "Spring AI",
        "Claude Code",
        "Hybrid RAG"
      ],
      "summary": "蚂蚁智能体与大模型应用一面，短而密集，命中幻觉治理、Skill 实现、grep 与 RAG 区别和混合召回。"
    },
    {
      "id": "nowcoder-alibaba-taobao-agent-2026-04-30",
      "platform": "牛客",
      "title": "阿里淘宝闪购 · Agent 算法工程师 · 27届实习一面",
      "company": "阿里淘宝闪购",
      "published_at": "2026-04-30",
      "source_url": "https://www.nowcoder.com/discuss/879393838081597440",
      "score": 5,
      "topics": [
        "Agent Framework",
        "HITL",
        "Risk Control",
        "Memory",
        "Token Cost",
        "Schema Validation",
        "AI Coding"
      ],
      "summary": "阿里淘宝闪购 Agent 算法实习面经，工程深挖质量高，适合补充 HITL、高风险 tool 管控和长周期记忆压缩。"
    },
    {
      "id": "nowcoder-tencent-baidu-agent-summary-2026-04-28",
      "platform": "牛客",
      "title": "大模型、Agent面经总结【04/28】腾讯 / 百度 总结",
      "company": "腾讯/百度",
      "published_at": "2026-04-28",
      "source_url": "https://www.nowcoder.com/discuss/878600528970735616",
      "score": 5,
      "topics": [
        "Agent Orchestration",
        "RAG Hot Update",
        "Failure Retry",
        "Financial Safety",
        "LoRA",
        "DPO",
        "Vector Retrieval"
      ],
      "summary": "腾讯/百度多条近期面经聚合，覆盖技术、产品、算法多类岗位，适合拆分同步到公司面经。"
    },
    {
      "id": "nowcoder-ali-ant-byte-agent-summary-2026-04-24",
      "platform": "牛客",
      "title": "Agent 开发面经总结【04/24】阿里巴巴 / 蚂蚁 / 字节跳动 总结",
      "company": "阿里巴巴/蚂蚁/字节跳动",
      "published_at": "2026-04-24",
      "source_url": "https://www.nowcoder.com/discuss/877151327091027968",
      "score": 5,
      "topics": [
        "Multi-Agent",
        "RAG",
        "MCP",
        "Function Calling",
        "LangChain",
        "LangGraph",
        "SSE",
        "WebSocket"
      ],
      "summary": "多公司 Agent 开发聚合面经，覆盖 Agent 项目架构、多 Agent 通信、RAG 实现、工具调用和质量保障。"
    },
    {
      "id": "nowcoder-jd-agent-intern-2026-04-23",
      "platform": "牛客",
      "title": "京东 Agent开发 暑期一面",
      "company": "京东",
      "published_at": "2026-04-23",
      "source_url": "https://www.nowcoder.com/discuss/876932752833077248",
      "score": 5,
      "topics": [
        "Agent Workflow",
        "E-commerce RAG",
        "Intent Recognition",
        "Redis Vector Search",
        "HNSW",
        "Hallucination Fallback"
      ],
      "summary": "京东 Agent 暑期一面，RAG 工程细节丰富，适合补充电商知识库、意图识别评测和 Redis 向量索引。"
    },
    {
      "id": "nowcoder-ant-international-2025-09-24",
      "platform": "牛客",
      "title": "蚂蚁秋招时间线+面经",
      "company": "蚂蚁国际",
      "published_at": "2025-09-24",
      "source_url": "https://www.nowcoder.com/discuss/800426409796624384",
      "score": 5,
      "topics": [
        "RAG vs Fine-tuning",
        "Rerank",
        "NDCG",
        "Agent Evaluation",
        "Memory",
        "MCP",
        "A2A"
      ],
      "summary": "蚂蚁国际秋招面经，RAG 评测问题集中，适合补充 RAG 与微调取舍、Agent 效果评测、MCP/A2A 通信。"
    },
    {
      "id": "nowcoder-bytedance-llm-offer-2025-04-28",
      "platform": "牛客",
      "title": "4轮拿下字节Offer！LLM面试题合集",
      "company": "字节跳动",
      "published_at": "2025-04-28",
      "source_url": "https://www.nowcoder.com/discuss/746382064101908480",
      "score": 5,
      "topics": [
        "RAG Ranking",
        "Prompt Evaluation",
        "ReAct",
        "LoRA",
        "SFT",
        "Attention",
        "Hallucination"
      ],
      "summary": "字节多轮 LLM 面经，覆盖 RAG 链路、Prompt 评测、LoRA/SFT、幻觉和 token 限制处理。"
    },
    {
      "id": "nowcoder-bytedance-ai-app",
      "platform": "牛客",
      "title": "字节 AI 应用岗面试真题",
      "company": "字节跳动",
      "published_at": null,
      "source_url": "https://www.nowcoder.com/discuss/882634966025175040",
      "score": 5,
      "topics": [
        "Legal RAG",
        "Tool Agent",
        "Coze",
        "Doubao",
        "Evaluation",
        "Tool Routing",
        "Chunking",
        "Rerank"
      ],
      "summary": "字节 AI 应用岗深度复盘型面经，适合补充法律 RAG、工具路由、Recall@5 评估和文本切分。"
    },
    {
      "id": "zhihu-meituan-llm-intern",
      "platform": "知乎",
      "title": "Meituan Large Model Algorithm Intern questions",
      "company": "美团",
      "published_at": null,
      "source_url": "https://www.zhihu.com/en/article/688624199",
      "score": 5,
      "topics": [
        "RAG Project",
        "Metrics",
        "Qwen",
        "LoRA",
        "PDF Parsing",
        "Table Parsing"
      ],
      "summary": "知乎公开英文页，美团大模型算法实习面试风格强，适合补充 RAG 项目架构、评估、坏召回处理和表格解析。"
    },
    {
      "id": "zhihu-li-auto-llm-intern",
      "platform": "知乎",
      "title": "Li Auto Large Model Algorithm Intern experience",
      "company": "理想汽车",
      "published_at": null,
      "source_url": "https://www.zhihu.com/en/article/680860432",
      "score": 5,
      "topics": [
        "RAG",
        "Dataset Scale",
        "SFT",
        "CoT",
        "ToT",
        "Deployment",
        "vLLM"
      ],
      "summary": "理想汽车大模型算法实习面经，问题围绕完整 RAG I/O、部署效率、上线后问题、SFT 与推理框架。"
    },
    {
      "id": "csdn-agent-203-interviews-2026-06-18",
      "platform": "CSDN",
      "title": "AI Agent开发面试高频题曝光！从203篇面经提炼",
      "company": null,
      "published_at": "2026-06-18",
      "source_url": "https://blog.csdn.net/Trb701012/article/details/162102884",
      "score": 5,
      "topics": [
        "High Frequency Questions",
        "RAG",
        "Agent",
        "Engineering"
      ],
      "summary": "从大量面经中提炼 AI Agent 应用开发高频问题，适合作为专题题库种子和优先级参考。"
    },
    {
      "id": "cnblogs-byte-feilian-agent-2026-05-26",
      "platform": "博客园",
      "title": "字节 AI Agent 二面（飞连）面试题与参考解答",
      "company": "字节跳动",
      "published_at": "2026-05-26",
      "source_url": "https://www.cnblogs.com/tuaran/p/20164742",
      "score": 5,
      "topics": [
        "Tool Calling",
        "Bad Case",
        "Memory",
        "RAG Optimization",
        "Agentic RAG",
        "LangGraph"
      ],
      "summary": "字节飞连 Agent 二面题与参考解答，适合补充字节公司文档和 Agentic RAG 专题。"
    },
    {
      "id": "csdn-byte-agent-memory-rag",
      "platform": "CSDN",
      "title": "字节跳动大模型实习面经：从 Agent 记忆到 RAG 优化",
      "company": "字节跳动",
      "published_at": null,
      "source_url": "https://gitcode.csdn.net/6a2ccdca10ee7a33f27c07bf.html",
      "score": 5,
      "topics": [
        "Memory",
        "Query Rewrite",
        "Hybrid Retrieval",
        "RRF",
        "Rerank",
        "HyDE",
        "vLLM",
        "SGLang"
      ],
      "summary": "字节大模型应用算法实习候选，工程细节强，适合补充记忆系统、RAG 优化和推理框架追问。"
    },
    {
      "id": "csdn-bytedance-feishu-rag",
      "platform": "CSDN",
      "title": "双非本｜字节跳动飞书团队 RAG 面经",
      "company": "字节跳动",
      "published_at": null,
      "source_url": "https://devpress.csdn.net/v1/article/detail/151567056",
      "score": 5,
      "topics": [
        "BGE-M3",
        "Qwen3-Embedding",
        "LoRA",
        "Hybrid Retrieval",
        "RRF",
        "Parent-child Documents"
      ],
      "summary": "字节飞书 RAG 面经，适合补充 embedding 选型、多路召回、chunk overlap 和父子文档。"
    },
    {
      "id": "csdn-22-company-llm-app-interviews",
      "platform": "CSDN",
      "title": "大模型应用开发面试宝典：22家公司真实面试经验与技术考点总结",
      "company": null,
      "published_at": null,
      "source_url": "https://devpress.csdn.net/v1/article/detail/151832648",
      "score": 5,
      "topics": [
        "LLM",
        "RAG",
        "Agent",
        "MCP",
        "System Design"
      ],
      "summary": "覆盖阿里、腾讯、字节等 LLM 应用岗的面试经验与考点总结，适合做多公司题源索引。"
    },
    {
      "id": "juejin-rag-interview-2026-06-18",
      "platform": "掘金",
      "title": "RAG大厂面试题汇总：向量检索、混合检索、Rerank、幻觉处理",
      "company": null,
      "published_at": "2026-06-18",
      "source_url": "https://juejin.cn/post/7652557621874409522",
      "score": 5,
      "topics": [
        "RAG",
        "Vector Search",
        "BM25",
        "RRF",
        "Rerank",
        "Agentic RAG",
        "Hallucination"
      ],
      "summary": "RAG 专项高质量题源，适合并入 RAG 核心知识或最新面经索引。"
    },
    {
      "id": "juejin-agent-offers-2026-04-07",
      "platform": "掘金",
      "title": "2026 最新 AI Agent 岗面试复盘：拿到三个 offer",
      "company": null,
      "published_at": "2026-04-07",
      "source_url": "https://juejin.cn/post/7625576464485842979",
      "score": 5,
      "topics": [
        "Framework Selection",
        "Failure Modes",
        "ReAct",
        "Tree of Thoughts",
        "Customer Service Agent"
      ],
      "summary": "AI Agent 岗复盘型资料，适合整理为准备策略和企业客服 Agent 系统设计。"
    },
    {
      "id": "cnblogs-agent-40-questions-2026-05-21",
      "platform": "博客园",
      "title": "面试 AI Agent 工程师会被问什么？40+ 真题 + 知识图谱全梳理",
      "company": null,
      "published_at": "2026-05-21",
      "source_url": "https://www.cnblogs.com/itech/p/20111938",
      "score": 5,
      "topics": [
        "ReAct",
        "Memory",
        "Function Calling",
        "MCP",
        "A2A",
        "RAG",
        "Evaluation"
      ],
      "summary": "知识图谱型 Agent 面试题整理，适合补充通用最新面经索引。"
    },
    {
      "id": "github-agentguide",
      "platform": "GitHub",
      "title": "AgentGuide",
      "company": null,
      "published_at": "2026-06-29",
      "source_url": "https://github.com/adongwanai/AgentGuide",
      "score": 5,
      "topics": [
        "AI Agent",
        "LangGraph",
        "Advanced RAG",
        "LLM Interview"
      ],
      "summary": "AI Agent 开发指南，覆盖 LangGraph、高级 RAG、大模型面试，适合作为外部高质量资料索引。"
    },
    {
      "id": "github-ai-agent-interview-guide",
      "platform": "GitHub",
      "title": "ai-agent-interview-guide",
      "company": null,
      "published_at": "2026-06-29",
      "source_url": "https://github.com/bcefghj/ai-agent-interview-guide",
      "score": 5,
      "topics": [
        "AI Agent Interview",
        "Project",
        "Resume",
        "STAR",
        "System Design",
        "Python",
        "Java",
        "Go"
      ],
      "summary": "AI Agent 面试全攻略，包含题库、项目、简历模板和系统设计，适合作为参考资料索引。",
      "source_type": "reference",
      "confidence": "high",
      "verified_at": "2026-09-11"
    },
    {
      "id": "github-llmforeverybody",
      "platform": "GitHub",
      "title": "LLMForEverybody",
      "company": null,
      "published_at": "2026-06-29",
      "source_url": "https://github.com/luhengshiwo/LLMForEverybody",
      "score": 5,
      "topics": [
        "LLM",
        "Paper Roadmap",
        "Interview Bank",
        "Agent",
        "RAG"
      ],
      "summary": "大模型知识与面试准备资源，适合作为理论学习和面试准备索引。"
    },
    {
      "id": "github-faq-llm-interview",
      "platform": "GitHub",
      "title": "FAQ_Of_LLM_Interview",
      "company": null,
      "published_at": "2026-06-29",
      "source_url": "https://github.com/aceliuchanghong/FAQ_Of_LLM_Interview",
      "score": 5,
      "topics": [
        "LLM Interview",
        "Algorithm Role",
        "Q&A"
      ],
      "summary": "大模型算法岗面试题含答案，适合作为基础八股和算法岗准备资料索引。"
    },
    {
      "id": "github-lau-llm-agent-guide",
      "platform": "GitHub",
      "title": "Lau-Jonathan/LLM-Agent-Interview-Guide",
      "company": null,
      "published_at": "2026-02",
      "source_url": "https://github.com/Lau-Jonathan/LLM-Agent-Interview-Guide",
      "score": 5,
      "topics": [
        "Transformer",
        "Inference Optimization",
        "Fine-tuning",
        "RAG",
        "Agent",
        "Safety",
        "System Design"
      ],
      "summary": "300+ Q&A 与字节/阿里/腾讯真题，适合做外部题库参考和查漏补缺。"
    },
    {
      "id": "github-anglemaxin-llm-app-interview",
      "platform": "GitHub",
      "title": "AngleMAXIN/llm-application-interview",
      "company": null,
      "published_at": null,
      "source_url": "https://github.com/AngleMAXIN/llm-application-interview",
      "score": 5,
      "topics": [
        "Real Interviews",
        "LLM Application",
        "RAG",
        "Agent",
        "Workflow",
        "System Design"
      ],
      "summary": "多家大厂真实 LLM 应用面试问题，适合从中抽取公司题和通用项目深挖题。"
    },
    {
      "id": "github-javaguide-ai",
      "platform": "GitHub",
      "title": "JavaGuide docs/ai",
      "company": null,
      "published_at": null,
      "source_url": "https://github.com/Snailclimb/JavaGuide/tree/main/docs/ai",
      "score": 5,
      "topics": [
        "AI Agent",
        "LLM Basics",
        "RAG",
        "AI System Design"
      ],
      "summary": "高可信中文开发者资源，适合作为 AI Agent、RAG、系统设计专题参考。"
    },
    {
      "id": "github-tobebetterjavaer-ai",
      "platform": "GitHub",
      "title": "toBeBetterJavaer AI section",
      "company": null,
      "published_at": null,
      "source_url": "https://github.com/itwanger/toBeBetterJavaer/tree/master/docs/src/sidebar/itwanger/ai",
      "score": 5,
      "topics": [
        "Agent 258 Questions",
        "LLM 333 Questions",
        "RAG",
        "MCP",
        "Interview Walkthrough"
      ],
      "summary": "Agent 258 题、大模型 333 题和 RAG+Agent+MCP 面试 walkthrough，适合作为外部题库资源。"
    },
    {
      "id": "github-wdndev-llm-interview-note",
      "platform": "GitHub",
      "title": "wdndev/llm_interview_note",
      "company": null,
      "published_at": null,
      "source_url": "https://github.com/wdndev/llm_interview_note",
      "score": 5,
      "topics": [
        "Transformer",
        "SFT",
        "LoRA",
        "RAG",
        "Agent",
        "Inference",
        "Evaluation"
      ],
      "summary": "中文 LLM 面试笔记，覆盖 RAG、Agent、推理部署和评估，适合作为系统学习索引。"
    },
    {
      "id": "github-rag-interview-hub",
      "platform": "GitHub",
      "title": "RAG-Interview-Questions-and-Answers-Hub",
      "company": null,
      "published_at": null,
      "source_url": "https://github.com/KalyanKS-NLP/RAG-Interview-Questions-and-Answers-Hub",
      "score": 5,
      "topics": [
        "RAG",
        "Chunking",
        "HyDE",
        "GraphRAG",
        "RAGAS",
        "Failure Diagnosis"
      ],
      "summary": "英文 RAG 100+ Q&A，适合补充高级 RAG 题和英文参考。"
    },
    {
      "id": "nowcoder-xiaohongshu-agentic-fullstack-2026-08",
      "platform": "牛客",
      "title": "小红书大模型平台 Agentic 全栈研发练习生一面凉经",
      "company": "小红书",
      "published_at": "2026-08-07",
      "source_url": "https://www.nowcoder.com/feed/main/detail/e5e9311a623940eead6ec98c65e7f9e8",
      "score": 5,
      "topics": [
        "Agent Architecture",
        "Harness",
        "Skill",
        "LangGraph",
        "RAG",
        "Scalability"
      ],
      "summary": "一手凉经，集中追问 Agent 编排、Skill/Tool、模型一致性、十几万次日请求扩容、SQLite 选型和 LangGraph/自研框架取舍。"
    },
    {
      "id": "nowcoder-mihoyo-agent-first-round-2026-08",
      "platform": "牛客",
      "title": "米哈游 AI Agent 开发提前批一面面经分享",
      "company": "米哈游",
      "published_at": "2026-08-09",
      "source_url": "https://www.nowcoder.com/discuss/916476300372500480",
      "score": 4,
      "topics": [
        "AI Coding",
        "Hallucination",
        "MCP",
        "Skill",
        "Memory",
        "Agent Evaluation"
      ],
      "summary": "近期答题解析型面经，覆盖 AI coding 质量门禁、幻觉治理、Skill/MCP、Agent 项目深挖和现场编码。"
    },
    {
      "id": "nowcoder-taotian-agent-social-first-round-2026-07",
      "platform": "牛客",
      "title": "淘天 Agent 社招一面面经分享",
      "company": "阿里/淘天",
      "published_at": "2026-07-22",
      "source_url": "https://www.nowcoder.com/discuss/909920471301226496",
      "score": 4,
      "topics": [
        "Multi-task Learning",
        "DPO",
        "GRPO",
        "Memory",
        "Skill",
        "Trajectory Reward"
      ],
      "summary": "近期社招答题解析，偏算法与训练，重点涉及多任务冲突、DPO/GRPO、Agent 定义、Memory 和多步工具轨迹奖励。"
    },
    {
      "id": "nowcoder-bytedance-ant-tencent-agent-intern-2026-07",
      "platform": "牛客",
      "title": "字节、蚂蚁、腾讯 Agent 实习面经",
      "company": "字节跳动/蚂蚁/腾讯",
      "published_at": "2026-07-07",
      "source_url": "https://www.nowcoder.com/discuss/904029765160497152",
      "score": 5,
      "topics": [
        "Transformer",
        "RAG",
        "LoRA",
        "Model Training",
        "Algorithms"
      ],
      "summary": "一手多公司凉经，展示字节偏基础与项目、腾讯调剂岗位偏算法、蚂蚁偏微调与 RAG 的考察差异。"
    },
    {
      "id": "blanked-openai-anthropic-interviews-2026-08",
      "platform": "Blanked",
      "title": "28 份 OpenAI 与 Anthropic 面试经历对比",
      "company": "OpenAI/Anthropic",
      "published_at": "2026-08-06",
      "source_url": "https://www.blanked.work/blog/openai-vs-anthropic-interviews",
      "score": 5,
      "topics": [
        "System Design",
        "Reliability",
        "Safety",
        "Production Engineering",
        "Behavioral"
      ],
      "summary": "基于 28 份公开候选人经历的研究：OpenAI 更频繁追问规模、失败模式与可靠性，Anthropic 更强调安全判断和不确定条件下的决策。"
    },
    {
      "id": "nowcoder-alibaba-taotian-agent-intern-2026-04",
      "platform": "牛客",
      "title": "阿里淘天 Agent 开发日常实习一面",
      "company": "阿里/淘天",
      "published_at": "2026-04-15",
      "source_url": "https://www.nowcoder.com/feed/main/detail/566fa6594dcc446a82bd203f139c45c2",
      "score": 5,
      "topics": [
        "State Machine",
        "Workflow",
        "MCP",
        "Skill",
        "Context Engineering",
        "AI Coding"
      ],
      "summary": "一手实习面经，强烈关注 Agent/Workflow 边界、动态上下文、RAG 分块、ReAct/CoT 取舍和 AI coding 完整交付。"
    },
    {
      "id": "nowcoder-bytedance-agent-90min-2026-04",
      "platform": "牛客",
      "title": "字节 Agent 开发一面 90 分钟凉经",
      "company": "字节跳动",
      "published_at": "2026-04-09",
      "source_url": "https://www.nowcoder.com/feed/main/detail/91c5394e57c14927841d7a86bfe427c2",
      "score": 5,
      "topics": [
        "Code Agent",
        "Context Engineering",
        "Skill",
        "RAG",
        "Self-Attention",
        "Algorithms"
      ],
      "summary": "一手 90 分钟面经，项目追问覆盖代码 Agent、覆盖率与插桩、查询改写、上下文工程、Skills、LLM 原理和语言基础。"
    },
    {
      "id": "nowcoder-tencent-agent-app-2026-04",
      "platform": "牛客",
      "title": "腾讯 AI Agent 应用开发一面凉经",
      "company": "腾讯",
      "published_at": "2026-04-04",
      "source_url": "https://www.nowcoder.com/feed/main/detail/28edcddf0c204c08b6562a3e6e6b73ae",
      "score": 5,
      "topics": [
        "Coding Agent",
        "SubAgent",
        "Context Engineering",
        "GRPO",
        "QLoRA",
        "Python"
      ],
      "summary": "一手凉经，围绕自研 Coding Agent 与 Claude Code 的差异持续追问，并延伸到 SubAgent、后训练和 Python 基础。"
    },
    {
      "id": "nowcoder-bytedance-agent-intern-2026-03",
      "platform": "牛客",
      "title": "字节 Agent 开发实习一面",
      "company": "字节跳动",
      "published_at": "2026-03-18",
      "source_url": "https://www.nowcoder.com/feed/main/detail/6506d4b4addf447c8e2c135b5088cdc8",
      "score": 5,
      "topics": [
        "Agent Framework",
        "Memory",
        "Context Compression",
        "RAG",
        "MCP",
        "A2A"
      ],
      "summary": "一手实习面经，集中覆盖 Memory、上下文压缩、RAG、MCP、Skill 渐进式披露、OpenClaw 与 A2A。"
    },
    {
      "id": "nowcoder-alibaba-taotian-agent-campus-2026-02",
      "platform": "牛客",
      "title": "阿里淘天大模型 Agent 校招面经",
      "company": "阿里/淘天",
      "published_at": "2026-02-27",
      "source_url": "https://www.nowcoder.com/feed/main/detail/78d6c8c30f1741e6b0a1a02d7b4bbfab",
      "score": 5,
      "topics": [
        "Attention",
        "SFT",
        "RAG Evaluation",
        "DPO",
        "GRPO",
        "Tool Orchestration"
      ],
      "summary": "一手校招面经，覆盖 Attention、SFT 与后训练、RAG 评估、Agent 多步规划、工具调度和延迟优化。"
    }
  ]
}