AGENT VIEW · Mokabla · Question · cron-scheduler-take-home
AgentGate · request ag_fb956f80b904

Who you are, as far as this app can tell

declared   ClaudeBot · Anthropic   · not verified

The user agent is ClaudeBot (Anthropic), which identifies itself as an AI agent.

Signals: Known agent user agent

This classification is final for the session. The human pages are not served to agents.

This page

page
question
human URL
https://www.mokabla.com/en/c/telda/cron-scheduler-take-home

State

question
object · slug, title, difficulty, tags, is_free, locked, frequency, last_reported_at, url, agent, company, body_md, follow_ups_md
paid
false
access
Question bodies are paid content: a person buys 30 days of access per company (200 EGP or 200 points) in the human app. Agents get titles, tags and difficulty; bodies only for free questions.
{
  "question": {
    "slug": "cron-scheduler-take-home",
    "title": "Cron Scheduler Take-Home",
    "difficulty": "hard",
    "tags": [
      "system-design",
      "oop-design",
      "swe",
      "backend",
      "take-home"
    ],
    "is_free": true,
    "locked": false,
    "frequency": 0.857,
    "last_reported_at": "2026-10-30T12:02:52.052164Z",
    "url": "https://www.mokabla.com/en/c/telda/cron-scheduler-take-home",
    "agent": {
      "page": "https://www.mokabla.com/agent/en/c/telda/cron-scheduler-take-home",
      "question": "https://www.mokabla.com/agent/_call/get_question?company=telda&question=cron-scheduler-take-home"
    },
    "company": {
      "slug": "telda",
      "name_en": "Telda"
    },
    "body_md": "## What they ask\nThe backend take-home has been the same brief since 2021, and the company keeps it in a public repository on its GitHub organisation. After the intro call an engineer sends you the link and a deadline, and the task comes back in the onsite loop as a dedicated discussion round.\n\nBuild an in-process cron scheduler that accepts a job and runs it periodically. A client registers a job with:\n- an expected duration for a single run, for example `30m`,\n- a frequency, for example `1hr` for a job that should run every hour,\n- the job itself, a function,\n- a unique job identifier.\n\nThe solution must run several jobs concurrently, ship with tests, and instrument every run (actual execution time, logging). It has to be written in Go, Java, Python, C++, Ruby or JavaScript, and it must not lean on a scheduling library. The brief names Java's `ScheduledExecutorService.scheduleAtFixedRate` as the kind of shortcut that is not allowed.\n\nThe README is graded as hard as the code. They ask for a short description, the reasoning behind your technical decisions, the trade-offs you made, usage snippets, and what you would improve next.\n\n## What candidates built\nPublished solutions cluster around three shapes: a priority queue keyed by next run time with worker threads, a plain loop per job with sleeps, and an actor-style design with a timer actor, a scheduler actor and one monitor per job. Deleting a job is a common pain point, and a tombstone set checked when the job reaches the head of the queue is the usual answer.\n\n## What they look for\n- A clear policy for the run that overruns its frequency: skip, queue, or run in parallel, and why.\n- What the expected duration is for. The good answer is a timeout and a metric, not a sleep.\n- Fixed rate versus fixed delay, and how you keep drift out of the schedule.\n- Thread safety when jobs are added or removed while the scheduler is running, and a graceful shutdown that lets in-flight runs finish.\n- Tests that do not sleep for real minutes: an injectable clock or a tiny tick granularity.\n- A README that reads like an engineer explaining a design to a colleague.",
    "follow_ups_md": "- A job's run takes longer than its frequency. What happens to the next run?\n- Stop one job while it is executing. What does the caller see?\n- How do you test an hourly job without waiting an hour?\n- Ten thousand jobs with different frequencies. Where does your design break first?"
  },
  "paid": false,
  "access": "Question bodies are paid content: a person buys 30 days of access per company (200 EGP or 200 points) in the human app. Agents get titles, tags and difficulty; bodies only for free questions."
}

Actions you may take here

actioninputpermission
get_question
One question with its body. Bodies are paid content unless the question is free.
company: string, question: stringAllowedcall →
list_questions
The questions reported at a company: titles, tags, difficulty, whether each is free or paid.
company: string, page: number?Allowedcall →
buy_access
Buy 30 days of access to a company.
company: stringRefused
R1 This action is for people only
people: https://www.mokabla.com/en/pricing
post_comment
Comment on a question.
question: string, text: stringRefused
R1 This action is for people only
people: https://www.mokabla.com/en/companies

Allowed actions answer in JSON. Add ?format=json or send Accept: application/json to get this whole page as data.

Navigate (pre-filled links you can follow)

Notices

This classification is final for the session: it came from a strong signal, not a guess. The human pages are not served to agents.
Unverified agents are read-only. Signed requests (Web Bot Auth) are the path to write access.