Palantir Software Engineer Phone Screen Questions

Phone Screen Same question as this https://leetcode.com/discuss/interview-question/4866657/2024-03-Palantir-Tech-Screening Proposed 2 pointer approach, follow up question was for if the query has multiple words, how would u do it. I told some...

Current Status: Undergraduate Student, Bachelor\'s in Computer Science and Engineering. Postion: Software Engineering Intern @ Palantir, London. Location: London, UK. I came to know about the opening from their careers page and applied...

## Problem Compress a string using run-length encoding but only when it reduces length, applying conditions for when to compress. ## Likely LeetCode equivalent Similar to String Compression (LC 443). ## Tags strings, two_pointers, palantir

## Problem Normalize email addresses by removing dots, plus-suffixes, or lowercasing to find duplicate accounts. ## Likely LeetCode equivalent Similar to Unique Email Addresses (LC 929). ## Tags strings, hash_table, palantir

## Problem Merge or query overlapping intervals representing planetary orbits or time windows. ## Likely LeetCode equivalent Similar to Merge Intervals (LC 56) or Insert Interval (LC 57). ## Tags sorting, arrays, palantir, intervals

## Problem You are building the content feed for a swipe-based app. Each user has a set of `seen_ids` (profiles already swiped) and a preference vector. Given a pool of candidate profiles with feature vectors, return the top `k` unseen candidates ranked by cosine similarity to the user's preference vector. Exclude already-seen profiles. ```python def recommend( user_pref: list[float], candidates: dict[str, list[float]], # id -> feature vector seen_ids: set[str], k: int ) -> list[str]: pass ``` ## Example ``` user_pref = [1.0, 0.0, 1.0] candidates = { "p1": [1.0, 0.0, 1.0], "p2": [0.0, 1.0, 0.0], "p3": [0.8, 0.1, 0.9], "p4": [1.0, 0.0, 0.5], } seen_ids = {"p2"} k = 2 Output: ["p1", "p3"] # cosine sim: p1=1.0, p3=~0.99, p4=~0.95; p2 excluded ``` ## Follow-ups 1. Cosine similarity is O(d) per candidate — how do you scale to millions of profiles? 2. How would you incorporate mutual-match signals (both users swiped right) into ranking? 3. How do you prevent a popularity bias where the same top profiles dominate all feeds?

## Problem Find the shortest distance between two given words in a text document, or match words within a distance threshold. ## Likely LeetCode equivalent Similar to Shortest Word Distance (LC 243). ## Tags strings, arrays, palantir

## Problem You have a table of web events: `(user_id, page_url, event_type, timestamp)`. A **session** is a sequence of events from the same user where no two consecutive events are more than 30 minutes apart. Compute: 1. Total sessions per user 2. Average session duration per user 3. Most visited page per session ```sql -- events table: user_id VARCHAR, page_url VARCHAR, event_type VARCHAR, ts TIMESTAMP -- Step 1: identify session boundaries SELECT user_id, ts, CASE WHEN ts - LAG(ts) OVER (PARTITION BY user_id ORDER BY ts) > INTERVAL '30 minutes' THEN 1 ELSE 0 END AS is_new_session FROM events; ``` ## Example ``` user_id | page_url | ts alice | /home | 10:00 alice | /shop | 10:15 <- same session (15 min gap) alice | /home | 11:00 <- new session (45 min gap) bob | /about | 10:05 Sessions: alice -> 2, bob -> 1 Alice session 1 duration: 15 min; session 2: 0 min ``` ## Follow-ups 1. How does this query change in a columnar store like BigQuery vs. a row store like Postgres? 2. How would you define and detect bot sessions? 3. If the events table has 10 billion rows, how do you run this efficiently?