GEP Worldwide-Technology Intern

GEP Worldwide | Technology Intern (2 Months) | On-Campus | Offer

• Status: Selected / Offer Received
• Position: Technology Intern (2 Months)
• Location: India
• Rounds: Online Assessment + 2 Technical Rounds + HR Discussion
• Timeline: ~3 days from OA to Offer
• Original Post / Reference Link: https://www.nexus-svnit.in/interview-experiences/post/6a9a6bfd18ad45131d1c87cf

---

Round 0: Online Assessment (OA)

• Duration: ~60 Minutes
• Platform: Online Test Platform

Sections:

1. Quantitative & Logical Aptitude: Standard numerical reasoning questions with an emphasis on Time & Work problems.
2. CS Fundamentals: MCQs covering OOP concepts, Database Normalization, Indexing, and standard SQL queries.
3. Coding Question (1 Problem):
   • Problem: Missing Number (Given an array of size n-1 containing numbers from 1 to n, find the single missing number).
   • Solution Discussion:
     • Sorting: $O(N \log N)$ time, $O(1)$ space
     • Hash Set: $O(N)$ time, $O(N)$ space
     • Optimal Math Approach: $O(N)$ time, $O(1)$ space using $\frac{n(n+1)}{2} - \sum \text{nums}$
     • Optimal Bitwise Approach: $O(N)$ time, $O(1)$ space using XOR to prevent integer overflow:

       int missingNumber(vector<int>& nums) {
           int n = nums.size() + 1;
           int xorVal = 0;
           for (int i = 1; i <= n; i++) xorVal ^= i;
           for (int x : nums) xorVal ^= x;
           return xorVal;
       }
       

---

Round 1: Technical Interview (15 Minutes)

• Mode: MS Teams + Notepad (No IDE / Compiler)

1. Introduction & Domain Scoping

• My background is in Artificial Intelligence. The interviewer immediately asked whether my work focuses on Foundational AI (mathematics, loss functions, pre-training) or Applied AI (fine-tuning, application engineering, RAG pipelines).
• Clarified that my focus is on Applied AI, directing the technical discussion toward production architectures.

2. Applied AI & LLM Systems

• RAG Architecture: High-level pipeline, vector embeddings, chunking, and handling hallucination.
• Vector Stores & ChromaDB: Embeddings generation, vector indexing, distance metrics (Cosine Similarity vs. Euclidean Distance).

3. Live Coding on Notepad: Remove Duplicates from Sorted Array

• Problem: Given a sorted vector containing duplicates (e.g., [1, 1, 1, 2, 2, 3, 3]), modify it in-place so each element appears only once.

• Approaches Discussed:

  • std::set: $O(N)$ auxiliary space
  • std::unordered_map: $O(N)$ auxiliary space
  • Two Pointers ($O(N)$ time, $O(1)$ space): Keep slow pointer i at unique boundaries, scan fast pointer j, assign arr[++i] = arr[j].

• C++ STL Follow-up:

  • The interviewer asked about the built-in C++ method for this.
  • Answer: std::unique from <algorithm> (Erase-Remove Idiom):

    #include <vector>
    #include <algorithm>
    
    vector<int> arr = {1, 1, 1, 2, 2, 3, 3};
    auto it = unique(arr.begin(), arr.end());
    arr.erase(it, arr.end()); // arr is now [1, 2, 3]
    

• Verdict: Advanced to Round 2. Advised to stay relaxed and keep steady composure.

---

Round 2: Technical & Leadership Round (20 Minutes)

• Interviewer: Principal Software Engineer (12+ years at GEP)
• Format: High-intensity rapid-fire verbal round (30+ questions in ~15 minutes) with zero latency to test spontaneous recall.

1. Prior Round Verification

• Interviewer asked who took my Round 1 and had me summarize every single topic and problem discussed.

2. DBMS Fundamentals

• Difference between WHERE and HAVING clauses.
• An advanced/niche database concept: I had not encountered it, so I stated clearly: "Sir, I have not worked with this particular concept yet." The interviewer appreciated the directness and moved on.

3. Rapid-Fire OOP (30+ Questions)

• 4 Pillars: Real-world examples and design trade-offs of Abstraction, Encapsulation, Inheritance, and Polymorphism.
• Encapsulation: Real-world analogies and code patterns (private data members exposed via validated getters/setters).
• Constructors: Types (Default, Parameterized, Copy), declaration syntax, scopes, and declaration inside classes vs. interfaces.
• Destructors: Syntax (~), execution lifecycle, and whether destructors can be overloaded (No, only one destructor per class).
• Polymorphism & Binding: Compile-time (function/operator overloading) vs. Runtime (virtual functions/overriding). Early/Static binding vs. Late/Dynamic binding via vtable and vptr.
• Interfaces & Abstract Classes: Pure virtual function syntax (= 0) and runtime polymorphism.

4. Data Structures (Verbal Walkthrough)

• Singly Linked List: Memory layout vs. Contiguous Arrays.
• Reverse a Linked List: Walked through the in-place 3-pointer iterative reversal (prev, curr, next) in $O(N)$ time and $O(1)$ space.

5. Questions for Interviewer

• Asked about his team's tech stack and engineering architecture.
• Asked how technical decision-making and architecture evolve across a 12-year tenure at GEP.

---

Round 3: HR Discussion (10 Minutes)

• Family background and educational journey.
• Confirmation of relocation preferences and readiness for the 2-month summer internship duration.
• Document and identity verification check.

---

Key Takeaways

1. Be Decisive When You Don't Know: During rapid-fire rounds, guessing wastes time and hurts credibility. If you don't know a concept, say so within 2 seconds. Senior engineers value honesty over guessing.
2. Practice Plain-Text Coding: Practice coding without IDE autocompletion or compiler error messages. Writing clean code in Notepad reveals whether your syntax and pointer mechanics are intuitive.
3. Know Your Resume Boundaries: If you have specialized domains like AI/ML on your resume, clearly delineate between foundational theory and applied engineering to keep the conversation within your strengths.
4. Learn STL Under-the-Hood: Along with standard algorithmic techniques (two-pointer, sliding window), know the corresponding STL implementations (e.g., std::unique, std::lower_bound).
5. Composure: Getting nervous is normal, but do not let it accelerate your speech. Take a one-second pause before speaking, structure your answer chronologically, and treat the interview as a collaborative engineering discussion.

---

• Campus Archive / Nexus Link: https://www.nexus-svnit.in/interview-experiences/post/6a9a6bfd18ad45131d1c87cf