Palantir · Phone screen → Onsite
In processPalantir FDSE Experienced-Hire Interviews: 5 Experiences
Palantir FDSE experienced-hire interviews in 2024–2025 centered on open-ended decomposition designs, short coding and learning an internal query language.
At a glance
- Process: recruiter call → technical screen (about 20 minutes of design, then a short coding task) → onsite with decomposition, learning and coding rounds (entries 01–04).
- Asked recently: an open-ended design in all 4 current reports, such as medicine distribution in a poor region and an app connecting coworkers with shared interests; an internal SQL-like language in 2 of 4 (entries 01, 04).
- Difficulty: the coding was called simple or about LeetCode Easy (entries 02, 03, 04); candidates stressed clarifying the problem and talking it through rather than the algorithm.
- Practise first: Relational Learning Queries, Shared-Interest Connection App and Employees With Shared Interests.
Coverage: 4 current reports (September 2024–July 2025) and 1 from 2021 in History. All five are Forward Deployed Software Engineer (FDSE) candidates changing jobs; one names the New York Foundry team.
What came up
Every current candidate got an open-ended design with the problem left loose on purpose; the skill being tested was shaping it with the interviewer. Two also had to learn an internal SQL-like language on the spot, under different round names.
| Topic | Reports | Last seen |
|---|---|---|
| Open-ended product or system design | 4 of 4 | Jul 2025 (report date) |
| Short coding exercise | 3 of 4 | Jul 2025 (report date) |
| Internal SQL-like language | 2 of 4 | Jul 2025 (report date) |
| Why Palantir | 1 of 4 | May 2025 (report date) |
Questions we found
System Design / Decomposition
| Question | Reports | Last seen |
|---|---|---|
| Medicine distribution in a poor region (practice) | 1 | Jul 2025 (report date) |
| App connecting coworkers who share interests | 1 | Jul 2025 (report date) |
| Chat for people with mutual interests (practice) | 1 | May 2025 (report date) |
| Spotify/SoundCloud-style recommendations (practice) | 1 | Sep 2024 (report date) |
| Reduce driver delay in an Uber-like service (practice) | 1 | Sep 2024 (report date) |
Algorithms / Data Structures
| Question | Reports | Last seen |
|---|---|---|
| Employees With Shared Interests | 1 | Jul 2025 (report date) |
| A simple stock-trading task (related practice) | 1 | Sep 2024 (report date) |
SQL-like learning round
| Question | Reports | Last seen |
|---|---|---|
| Learn an internal language, then join and aggregate tables (practice) | 1 | Sep 2024 (report date) |
| Simple questions in an older internal SQL-like language (not shared) | 1 | Jul 2025 (report date) |
Behavioral (BQ)
| Question | Reports | Last seen |
|---|---|---|
| What do you know about Palantir, and why do you want to join? | 1 | May 2025 (report date) |
4 detailed candidate timelines
01: medicine distribution, with the problem left open
Report date: July 25, 2025 · Role: FDSE · Result: Rejected
The post first read like a technical screen; the candidate later clarified these were two onsite rounds.
Learning
- Work in an older internal SQL-like language. The candidate found it fairly simple and stressed talking with the interviewer. The queries, tables and outputs were not shared.
Decomposition
- Build a system to distribute medicine in a poor region. Which problem to solve, and how to define it, was left for the candidate to explore with the interviewer's guidance. It only lightly touched classic system design; the candidate's advice was to ask questions, pin down details and avoid overengineering. The closest practice is the Medicine Delivery Platform, which adds prescription, pharmacy, inventory and payment rules the candidate did not mention.
The candidate was rejected without a reason and felt the judgment was subjective.
02: a shared-interest app, then employee matching
Report date: July 14, 2025 · Role: FDSE · Round: 45-minute decomposition, design and coding · Result: Not reported beyond reaching this round
After the HR round
- Design an app that connects coworkers who share interests.
- Coding in the same scenario: given employees and their IDs, find people with shared interests. The candidate called it about LeetCode Easy. The Employees With Shared Interests practice version spells out the record layout, matching rule and output order, which the post did not.
03: a recruiter call and a 20-minute chat design
Report date: May 19, 2025 · Role: Forward Deployed Software Engineer · Applied: online · Result: Failed (stage not stated)
Recruiter call (30 minutes, about two weeks after applying)
- What do you know about Palantir, why do you like it, and why do you want to join?
Technical screen (HackerRank)
- About 20 minutes on software that lets people with mutual interests chat. The prompt was open-ended and did not define "mutual interest". Closest practice: One-to-One Chat Service.
- About 20 minutes on a simple LeetCode-style problem the candidate did not name.
04: recommendations, an internal query language and driver delay
Report date: September 16, 2024 (the post header lists April–June 2023) · Role: Forward Deployed Software Engineer · Team: Foundry, New York · Result: Failed
Technical screen
- Design a Spotify- or SoundCloud-style recommendation service, including table schemas and APIs. Closest practice: Personalized Recommendation System; the candidate gave no ranking model or scale targets.
Onsite
- Decomp: learn an internal SQL-like language that can create tables, join them and aggregate rows, then answer simple questions with it. Practice: Relational Learning Queries. The candidate's tip was to clarify the syntax before starting.
- A high-level design, without diagrams, of an Uber-like service aimed at reducing driver delay. Closest practice: Ride-Hailing Dispatch and Trip Lifecycle, which adds dispatch rules not in this report.
- A simple coding task. The candidate could not find the identical problem and pointed to Best Time to Buy and Sell Stock II as related; practice it as Maximum Profit with Repeated Stock Trades.
The candidate felt clear spoken communication mattered most in these rounds.
Patterns across the reports
- Design and coding in one short screen. Entries 02 and 03 both moved from a short design into a coding task in the same session; entry 02's coding reused the design's scenario.
- Round names don't match the work. Entry 01 called its query-language round "Learning"; entry 04 called its SQL-like round "Decomp". Both meant learning an internal query language.
Common questions
What is the Palantir decomposition round?
An open-ended problem, such as distributing medicine in a poor region, where you define the problem with the interviewer before designing anything (entries 01, 04). Candidates said it is lighter on classic system design than expected.
What is the Palantir learning round?
You learn an internal SQL-like language during the interview and use it to join and aggregate tables (entries 01, 04).
How hard is Palantir FDSE coding?
Simple: about LeetCode Easy in entry 02, and "simple" in entries 03 and 04.
How long is the Palantir FDSE technical screen?
About 40 minutes in entry 03: roughly 20 minutes of design and 20 of coding. Entry 02's combined round was 45 minutes.
History (before 2023)
H01: a road-congestion problem that had to be "dynamic"
Report date: October 17, 2021 · Role: FDSE · Process: HR, then a technical interview · Result: Failed
- What data, metrics and algorithms would you use to ease congestion on a frequently jammed road? Pseudocode was enough. The candidate struggled and felt their direction differed from the interviewer's, and later recalled the answer had to be "dynamic", without saying what that changed.
Keep practicing
- Relational Learning Queries: the internal-language round in entry 04.
- Shared-Interest Connection App and Employees With Shared Interests: the design and coding in entry 02.
- Related design practice: Medicine Delivery Platform (entry 01), One-to-One Chat Service (entry 03), Personalized Recommendation System and Ride-Hailing Dispatch (entry 04); Repeated Stock Trades (entry 04).
- Palantir question banks: system design, algorithms and behavioral (sign-in needed).
- Everything Palantir in one place: the Palantir interview page and Palantir hiring insights.
- Other Palantir cohorts: internship · new grad, Part 1 · new grad, Part 2.
- More firsthand reports: the Experience feed.
Sources: Anywhere useful interview info can be found.
Comments
0No comments yet — be the first to share what you know.