Palantir · OA → Phone screen → Onsite
In processPalantir FDSE Karat and Learning: 7 Experiences (Part 2 of 2)
Palantir FDSE and FDE interviews, 2017–2022: seven intern and new grad loops in Sydney, London, Munich and New York with Karat, learning and design rounds.
At a glance
- Process (2017–2022): online test or recruiter screen → hour-long Karat interview (3 reports) → two or three hour-long onsite rounds of decomposition/design, learning from documentation, and coding → hiring manager.
- Asked in these reports: a multiplayer War simulation, a recipe app design with cookable-recipe selection, and a broken BFS over company ownership (all H15, 2019).
- Difficulty: one Karat ran easy, medium, then hard (H13); a Sydney candidate solved two of four learning tasks (H11).
- Practise first: the three 2019 London problems above; the current loops are in Part 1.
Coverage: 7 reports from 2017 to 2022: 5 FDSE/FDE internships (Sydney, London ×2, New York, one location not given) and 2 new grad roles (Munich, one location not given). All are more than three years old, so they sit in History below.
In this series
- Part 1: nine 2024–2026 accounts with short combined screens, a Government loop and the current question tables.
- Part 2 (current): seven 2017–2022 accounts with learning rounds, War, recipes and ownership graphs.
What came up
All seven loops paired design or decomposition with coding, and five had a learning round: extending code from documentation, learning a new language, or changing an existing codebase. Because every report here is older than three years, this part has no current tables; each question stays in its entry.
History (before 2023)
H10: new grad: a dataset design and documented modules
Interview date: late July to mid-August 2022 · Role: FDSE, new grad · Background: Master's student · Result: Not stated
- Onsite decomposition built around a sample dataset; the candidate stayed high-level, and the interviewer did not ask for low-level design.
- Learning round: extend supplied modules using library documentation, in Python, Java or TypeScript. The dataset, system and features were not named.
H11: Sydney internship, a Karat retake and four learning tasks
Report date: August 1, 2022 · Role: FDSE intern · Location: Sydney · Result: Rejected
- Five written application questions, then an OA (five-day deadline): three progressive tasks in 75 minutes, all solved.
- Hour-long Karat in Java: four questions on algorithms, complexity and file handling (10–15 minutes), then two problems coded and optimized. A retake went worse after interviewer-side technical problems; the candidate asked the recruiter to count the better attempt.
- A 30-minute recruiter call, then two hour-long interviews. Design: services, interactions, storage and queries, with hints after clarifying. Learning: four increasing tasks from documentation; two solved and the third explained but not coded. Behavioral: internships, projects, goals and situational questions.
- Rejected five days later; the task domains were not published.
H12: Munich new grad: codebase changes, then an offer
Interview date: before about March 2022 (report date June 15, 2022) · Role: FDE, new grad · Location: Munich · Result: Offer, accepted
- LinkedIn recruiter outreach; about two weeks later a video interview split between behavioral questions and LeetCode-style coding.
- Three hour-long onsites: a scalable application for a business problem, and two changes to an existing codebase using API documentation. Behavioral questions throughout, including past challenges and weaknesses.
- Positive feedback after a week. Before the hour-long hiring manager talk, the recruiter helped the candidate think through why Palantir and why FDE. Offer about a week later.
H13: London internship through the final round
Interview date: autumn 2021, for summer 2022 · Role: FDE intern · Location: London · Result: Rejected after the final interview
- One-hour Karat: about 15 minutes of pseudocode and time complexity, then easy, medium and hard coding.
- Two back-to-back hour-long interviews: learning an unfamiliar concept, and leading a decomposition discussion (neither topic was shared).
H14: a virtual superday and a professor-matching design
Interview date: fall 2020 · Role: FDSE intern · Result: Offer, accepted
- Online test, recruiter call, then an hour-long Karat: about 20 minutes on a side project and past work, then 40 minutes of coding with more problems until time ran out.
- Zoom superday, three hour-long rounds: coding, learning a new language from documentation, and system design. The approximate design example was matching professors and students with similar interests, across UI/UX, frontend, backend, schema and testing.
- A hiring manager interview followed.
H15: London: War, recipes, ownership and a surprise fourth round
Interview date: November 2019 · Role: FDSWE intern · Background: MEng student · Location: London · Result: Rejected
- Coding: multiplayer War with
nplayers and three cards face down during a war; not every helper was finished. - Design: an ingredient-based recipe app with recipe creation from ingredients and cook booking, then code the recipe-selection function.
- Debugging: a broken BFS over company ownership: find the owners of offshore companies, compute shares, then discuss cycles.
- An unexpected hiring manager round: about 20 minutes on the manager's work, five on the candidate, and 30 of coding (three hash maps and two nested BFSs; the prompt was not recalled). The first three rounds were rated hire or strong hire; the fourth led to a rejection.
H16: a New York onsite in 2017
Interview date: September 28, 2017 · Role: FDE intern · Location: New York · Result: Rejected
- Recruiter calls and a technical phone screen, then three onsite interviews in New York, including decomposition and a coding round where the code had to compile and run. The tasks were not shared.
Common questions
What is the Palantir Karat interview like?
An hour: 10–20 minutes of concepts, pseudocode or project talk, then coding that got harder until time ran out (H11, H13, H14).
What is the Palantir learning round?
You learn something new during the interview and use it: extend modules from library documentation (H10, H11), change an existing codebase from API docs (H12), or pick up a new language (H14).
Does Palantir ask debugging questions?
One 2019 loop had a broken BFS over company ownership to fix (H15).
Keep practicing
- From H15: Simulate Multiplayer War, Find Cookable Recipes, Trace Terminal Company Ownership and the recipe app design.
- Palantir question banks: algorithms and system design (sign-in needed).
- Everything Palantir in one place: the Palantir interview page and Palantir hiring insights.
- More firsthand reports: the Experience feed.
In this series
- Part 1
- Part 2 (current)
Sources: Anywhere useful interview info can be found.
Comments
0No comments yet — be the first to share what you know.