Palantir FDSE Interview Experience (REJECTION~)

Hello Everyone, I gathered a lot of resources while preparing for my Palantir interview. Although things didn’t go in my favor, I wanted to share my experience and insights from the process: # Intervi

Hello Everyone, I gathered a lot of resources while preparing for my Palantir interview. Although things didn’t go in my favor, I wanted to share my experience and insights from the process: # Interview Process: 1. 30-minute Behavioral Call – with Deployment Strategist 2. 1-hour Technical Interview (Leetcode - BFS/DFS) – Scheduled one week after the behavioral round 3. 1-hour Decomposition & Analytics Interview – 10 days after the technical round (I did not progress beyond this stage) I felt particularly strong in the Decomposition & Analytics portion, but ultimately, it didn’t work out. Here are some key takeaways from my experience: # Decomposition (20-25 min): * They asked me to design an application at a very high level. * I initially approached it as a standard system design interview, covering estimation, API logic, scalability, and security. However, the interviewer stopped me and redirected my focus to the functional aspects of the application instead. * I drew a data flow diagram and outlined API logic, but I went too deep into scalability and security (which I assumed would be critical for Palantir). # Analytics (Low-Level Design - 20-25 min): * Immediately after the Decomposition portion, I was given a few predefined classes based on what we had just discussed and was asked to implement a function with constraints. * I asked many clarifying questions regarding constraints and implementation. * I coded a working solution without much guidance, and optimization was not initially required it was asked as a follow-up after I finished coding. # Interview Questions: # 1. Decomposition: * Design a system where users can share their interests. # 2. Analytics Task: * Implement a function called search(user_id, users) Follw up questions * How would you optimize meeting history lookups? * Use a hashmap (key-value pair) for faster lookup instead of iterating through a list. * How would you optimize interest searches while keeping it as an array? * Sort interests alphabetically and use binary search on the first character to quickly find matching entries. * What additional functions would you implement to meet the requirements discussed in the Decomposition round? * Suggested adding user recommendations, filtering by interest categories, and an API to track user interactions. Hope this helps anyone preparing for the process!