Skip to main content
CodeSparring
Back to labs
Palantir· FDSEmedium60 min

911 Dispatch Optimization

The brief

You've been dropped into a city's 911 dispatch center. When an emergency call comes in, a human dispatcher stares at a live map of responder units (ambulances, fire, police) and manually picks who to send. Under load they pick slowly and inconsistently, and the GPS positions they're trusting can be seconds to minutes stale. Leadership wants software that recommends the best responder for each incident, but "best" is exactly what nobody has pinned down.

Your task

Across five milestones you'll scope what "best responder" actually means, decompose the dispatch workflow to find the real bottleneck, commit to a ranking contract that holds up under stale location data and concurrent assignment, then implement the recommender inside the existing dispatch codebase until the tests pass. Finally you'll defend your choices and grade yourself.

Why this company

The signature Palantir FDSE onsite round drops you into a vague, real-world operational problem and watches how you scope it. This lab mirrors that round: a messy 911 dispatch system where the constraints (stale GPS, concurrent assignment, who is allowed to see a caller's location, and a fail-safe fallback) matter far more than the ranking algorithm itself. It is one round of a longer loop. The real process also gates on a timed HackerRank assessment (a Python problem, a multi-table SQL query, and a REST endpoint you page through and aggregate) and a Learning round where you model an unfamiliar API cold, so treat this as onsite prep, not the whole bar.

What you'll do

  1. 1ClarifyDecomposition roundPin down what “best responder” means before you rank anything.
  2. 2DecomposeDecomposition roundMap the workflow, entities, and permissions, and the pieces most likely to break.
  3. 3DesignDecomposition roundCommit to a ranking contract and defend it under stale data and concurrency.
  4. 4BuildLearning / Re-engineering round (adjacent)Implement the recommender inside the real codebase and get the tests green.
  5. 5ReviewSelf-review + mission reflectionDefend your choices against the curveballs, then grade yourself.

What this prepares — and what to prep elsewhere

This lab rehearses

  • Decomposition round (the signature Palantir FDSE onsite round)
  • Learning / Re-engineering round (partial: you orient inside a real codebase)

Prep these elsewhere

Mode