{
  "checked_on": "2026-10-15",
  "scope": "Public interview reports and officially released exercises; not an exhaustive internet search or a verified question bank.",
  "reports": [
    {
      "id": "I01",
      "company": "Anthropic",
      "kind": "official",
      "role": "Performance Engineer",
      "date": "2026-01-21",
      "date_note": [
        "انتشار رسمی: ۲۱ ژانویهٔ ۲۰۲۶؛ تمرینِ قدیمی منتشر شده است.",
        "Official publication: 21 Jan 2026; a former take-home was released."
      ],
      "source": "Anthropic Engineering",
      "url": "https://www.anthropic.com/engineering/AI-resistant-technical-evaluations",
      "question": [
        "بهینه‌سازی برنامه روی یک شتاب‌دهندهٔ شبیه‌سازی‌شده؛ با اندازه‌گیری چرخه‌ها و تحلیل عملکرد.",
        "Optimize a program on a simulated accelerator using cycle measurements and performance analysis."
      ],
      "mapping": [
        "به فاز تکمیلی performance و serving مربوط است؛ این تمرینِ منتشرشده، سؤالِ طراحی کل پلتفرم ما نیست.",
        "Relevant to the later performance/serving track; this released exercise is not our entire platform design prompt."
      ],
      "anchors": [
        "gateway",
        "capacity"
      ],
      "limit": [
        "وقوع و انتشار این تمرین را خود شرکت تأیید کرده؛ سؤال فعلیِ هر تیم یا شباهت کامل با پروژه را تأیید نمی‌کند.",
        "The company confirms this former exercise and its release, not every team’s current prompt or a full match to our project."
      ],
      "short_title": [
        "بهینه‌سازی شتاب‌دهندهٔ شبیه‌سازی‌شده",
        "Simulated-accelerator optimization"
      ]
    },
    {
      "id": "I02",
      "company": "Anthropic",
      "kind": "candidate",
      "role": "Senior / Staff SWE · Inferencing",
      "date": null,
      "date_note": [
        "مصاحبه: ۲۰۲۵ طبق عنوان منبع؛ روز دقیق مشخص نیست.",
        "Interview: 2025 according to the source title; exact day unspecified."
      ],
      "source": "Aced / Exponent · candidate account",
      "url": "https://www.aced.io/experiences/anthropic-staff-software-engineer-interview-9bfb57",
      "question": [
        "طراحی API برای batch inference روی خوشهٔ GPU؛ تمرکزِ گزارش روی API، صف و زمان‌بندی پیرامون یک مدلِ جعبه‌سیاه است.",
        "Design a batch-inference API over GPU clusters; the account emphasizes the API, queueing, and scheduling around a black-box model."
      ],
      "mapping": [
        "Gateway، صف و ظرفیت؛ زمان‌بند چند GPU در scope نسخهٔ اول ما نیست.",
        "Gateway, queueing, and capacity; a multi-GPU scheduler is outside our first-version scope."
      ],
      "anchors": [
        "gateway",
        "capacity",
        "ADR-008"
      ],
      "limit": [
        "روایت داوطلب؛ انتساب سؤال به شرکت را مستقلاً تأیید نکرده‌ایم. برچسب Verified متعلق به خودِ ناشر است.",
        "A candidate account; company attribution is not independently verified. Any Verified badge belongs to the publisher."
      ],
      "short_title": [
        "طراحی API برای batch inference",
        "Batch-inference API design"
      ]
    },
    {
      "id": "I03",
      "company": "OpenAI",
      "kind": "candidate",
      "role": "Platform SWE",
      "date": null,
      "date_note": [
        "تاریخ دقیق مصاحبه در متن مشخص نیست.",
        "Exact interview date is not specified in the account."
      ],
      "source": "Reddit · self-reported experience",
      "url": "https://www.reddit.com/r/InterviewCoderHQ/comments/1rhfjpw/openai_swe_interview_experience_full_loop/",
      "question": [
        "ساخت سرویس تحویل webhook با retry، backoff و صف خطا؛ سپس بررسیِ گیرکردن تحویل پس از crash و راه‌حل مبتنی بر lease.",
        "Build reliable webhook delivery with retries, backoff, and a dead-letter queue; discuss stuck deliveries after a crash and lease-based recovery."
      ],
      "mapping": [
        "قرارداد delivery، بازیابی worker و idempotency در Runtime؛ این سؤال دربارهٔ RAG یا کل AI platform گزارش نشده است.",
        "Delivery contracts, worker recovery, and runtime idempotency; this report does not describe RAG or a complete AI platform prompt."
      ],
      "anchors": [
        "runtime",
        "failures",
        "ADR-005"
      ],
      "limit": [
        "روایتِ تأییدنشده در یک انجمن با زمینهٔ تبلیغ ابزار مصاحبه؛ اعتبار پایین‌تر از منبع رسمی است.",
        "Unverified self-report in a community that promotes an interview tool; weaker evidence than an official source."
      ],
      "short_title": [
        "تحویل webhook با retry و بازیابی crash",
        "Webhook delivery, retries, and crash recovery"
      ]
    },
    {
      "id": "I04",
      "company": "Databricks",
      "kind": "candidate",
      "role": "SWE L4 · Compute Platform",
      "date": null,
      "date_note": [
        "تاریخ دقیق مصاحبه در متن مشخص نیست.",
        "Exact interview date is not specified in the account."
      ],
      "source": "Reddit · self-reported experience",
      "url": "https://www.reddit.com/r/InterviewCoderHQ/comments/1rk933z/databricks_swe_interview_l4/",
      "question": [
        "زمان‌بند توزیع‌شدهٔ jobهای GPU؛ گزارش همچنین ساخت KV پایدار، WAL، بازیابی و مسئلهٔ برگشت worker قدیمی را شرح می‌دهد.",
        "A distributed GPU-job scheduler; the account also describes durable KV storage, WAL, recovery, and a returning stale worker."
      ],
      "mapping": [
        "Runtime، fencing و crash recovery؛ scheduling آموزش مدل یک مبحث مجاور است و در پروژهٔ اولیه پیاده نمی‌شود.",
        "Runtime, fencing, and crash recovery; training-job scheduling is adjacent material, not part of our initial implementation."
      ],
      "anchors": [
        "runtime",
        "failures",
        "ADR-015"
      ],
      "limit": [
        "همان نوع انجمن تبلیغاتی و روایت تأییدنشده؛ شاهد مستقل یا تضمینِ سؤال آینده نیست.",
        "The same promotional community context and unverified account; not independent corroboration or a future-question guarantee."
      ],
      "short_title": [
        "زمان‌بند GPU و دوام داده",
        "GPU scheduling and durable storage"
      ]
    },
    {
      "id": "I05",
      "company": "Intuit",
      "kind": "candidate",
      "role": "SWE-2",
      "date": null,
      "date_note": [
        "تاریخ دقیق مصاحبه در متن مشخص نیست.",
        "Exact interview date is not specified in the account."
      ],
      "source": "LeetCode Discuss · anonymous candidate",
      "url": "https://leetcode.com/discuss/post/7313117/intuit-interview-experience-for-swe-2-se-6t8a/",
      "question": [
        "افزودن API به مخزن Python/Flask با LLM کانتینری؛ سپس بحث تست، پایش و مفاهیم RAG، prompt و temperature.",
        "Add an API to a Python/Flask repository with a containerized LLM, then discuss testing, monitoring, RAG, prompts, and temperature."
      ],
      "mapping": [
        "اولین slice اجرایی، قرارداد API و تست سرتاسری؛ بخش RAG یک گفت‌وگوی مفهومی گزارش شده، نه طراحی کامل پلتفرم.",
        "The first runnable slice, API contracts, and end-to-end tests; RAG was reported as conceptual discussion, not a full platform design."
      ],
      "anchors": [
        "interfaces",
        "acceptance",
        "ADR-002"
      ],
      "limit": [
        "گزارشِ ناشناسِ داوطلب است؛ صحتِ هویت و وقوع مصاحبه مستقلاً تأیید نشده.",
        "An anonymous candidate report; identity and interview occurrence are not independently verified."
      ],
      "short_title": [
        "API یک LLM، تست و پایش",
        "An LLM API, tests, and monitoring"
      ]
    },
    {
      "id": "I06",
      "company": "NetApp",
      "kind": "candidate",
      "role": "SDE-1 · 1+ YOE",
      "date": null,
      "date_note": [
        "تاریخ دقیق مصاحبه در متن مشخص نیست.",
        "Exact interview date is not specified in the account."
      ],
      "source": "LeetCode Discuss · anonymous candidate",
      "url": "https://leetcode.com/discuss/post/8448835/netapp-sde-1-onsite-interview-by-anonymo-i7yv/",
      "question": [
        "بحث دربارهٔ pipelineهای RAG، vector database، retrieval و agent؛ همراه با مرور پروژه و هم‌زمانی و تراکنش‌ها.",
        "Discuss RAG pipelines, vector databases, retrieval, and agents alongside a project review, concurrency, and transactions."
      ],
      "mapping": [
        "Live Context و retrieval، سپس پیوند با تراکنش‌ها؛ منبع یک prompt مستقل برای ساخت AI platform ارائه نمی‌کند.",
        "Live Context and retrieval, connected to transactions; the source does not provide a standalone AI-platform design prompt."
      ],
      "anchors": [
        "retrieval",
        "data-model",
        "ADR-007"
      ],
      "limit": [
        "گزارش ناشناس از بحثِ مصاحبه؛ نه سند رسمیِ سؤال‌های شرکت.",
        "An anonymous account of interview discussion, not an official company question bank."
      ],
      "short_title": [
        "RAG و مرور سیستم‌های توزیع‌شده",
        "RAG and distributed-systems discussion"
      ]
    },
    {
      "id": "I07",
      "company": "Amazon",
      "kind": "candidate",
      "role": "SDE-1",
      "date": "2026-05-08",
      "date_note": [
        "مصاحبهٔ حضوری: ۸ مهٔ ۲۰۲۶ طبق timeline نویسنده.",
        "Onsite: 8 May 2026 according to the author’s timeline."
      ],
      "source": "LeetCode Discuss · anonymous candidate",
      "url": "https://leetcode.com/discuss/post/8471051/",
      "question": [
        "چطور در توسعه از LLM استفاده می‌کنی، خروجی تولیدشده را چگونه بررسی می‌کنی و چه وقت به آن اعتماد نمی‌کنی؟",
        "How do you use LLMs in development, verify generated output, and decide when not to trust it?"
      ],
      "mapping": [
        "شاهد و آزمون برای کیفیتِ خروجی و شیوهٔ کار؛ این بحث AI fluency است و سؤال system design محسوب نمی‌شود.",
        "Evidence and tests for output quality and engineering practice; this is AI-fluency discussion, not a system-design prompt."
      ],
      "anchors": [
        "evals",
        "acceptance",
        "ADR-011"
      ],
      "limit": [
        "روایت ناشناس؛ فقط همین گفت‌وگوی گزارش‌شده را پشتیبانی می‌کند، نه مصاحبهٔ نقش AI Engineer.",
        "An anonymous report supporting this discussion only, not a claim about an AI Engineer interview."
      ],
      "short_title": [
        "بررسی خروجی LLM در توسعه",
        "Checking LLM output during development"
      ]
    },
    {
      "id": "I08",
      "company": "Microsoft",
      "kind": "candidate",
      "role": "GBB Data & AI Specialist",
      "date": "2023-12-01",
      "date_note": [
        "تاریخ ثبت تجربه: ۱ دسامبر ۲۰۲۳؛ نمونهٔ قدیمی‌تر است.",
        "Experience date: 1 Dec 2023; an older example."
      ],
      "source": "Taro · candidate experience",
      "url": "https://www.jointaro.com/interviews/companies/microsoft/experiences/gbb-data-and-ai-specialist-united-states-december-1-2023-no-offer-positive-068d99e3/",
      "question": [
        "توضیح بده RAG برای LLM چگونه ساخته و استفاده می‌شود.",
        "Explain how to build and use RAG for an LLM."
      ],
      "mapping": [
        "مبانی retrieval و context؛ دربارهٔ نقش Specialist است و شاهدِ سؤال مهندسی پلتفرم در ۲۰۲۶ نیست.",
        "Retrieval/context fundamentals; this is a Specialist role and does not establish a 2026 platform-engineering prompt."
      ],
      "anchors": [
        "retrieval",
        "ADR-007"
      ],
      "limit": [
        "بازنشر تجربهٔ یک داوطلب؛ هویت و انتساب را مستقلاً تأیید نکرده‌ایم.",
        "A published candidate experience; identity and attribution are not independently verified."
      ],
      "short_title": [
        "توضیح نحوهٔ استفاده از RAG",
        "Explaining how to use RAG"
      ]
    }
  ],
  "industry_sources": [
    {
      "company": "Uber",
      "date": "2024-07-11",
      "url": "https://www.uber.com/us/en/blog/genai-gateway/",
      "title": "Navigating the LLM Landscape: Uber’s Innovation with GenAI Gateway",
      "topic": [
        "یک دروازهٔ مشترک برای دسترسی تیم‌ها به مدل‌ها، کنترل دسترسی، پایش و حساب‌کردن هزینه.",
        "A shared gateway for model access, authorization, monitoring, and cost attribution."
      ]
    },
    {
      "company": "Anthropic",
      "date": "2025-11-26",
      "url": "https://www.anthropic.com/engineering/effective-harnesses-for-long-running-agents",
      "title": "Effective harnesses for long-running agents",
      "topic": [
        "حفظِ پیشرفت و وضعیت برای agentهایی که کارشان از یک نشست و پنجرهٔ context طولانی‌تر می‌شود.",
        "Preserving progress and state for agents working across sessions and context windows."
      ]
    },
    {
      "company": "Anthropic",
      "date": "2026-01-09",
      "url": "https://www.anthropic.com/engineering/demystifying-evals-for-ai-agents",
      "title": "Demystifying evals for AI agents",
      "topic": [
        "ارزیابی چندمرحله‌ای با task، trial و grader؛ برای آشکارشدن تغییر کیفیت قبل از انتشار.",
        "Multi-step evaluation with tasks, trials, and graders to expose quality changes before release."
      ]
    }
  ]
}