InfoEdge (Naukri.com) - SDE Role | Interview Experience | 6M + FTE | OnCampus | 2026

Online Assessment - 1 hour

The OA consisted of:

• 3 basic DSA questions
• 35-45 MCQs covering: Verbal Ability, Statistics, Reasoning, AI-related concepts

Out of 473 eligible students, 23 were shortlisted after the OA.

Round 0 - (25 mins)

This was an AI-Interview round and non-eliminatory in nature. Total 8 questions including introduction. Generic questions on project, internship, SQL, OOPS, DBMS, version-control system etc.

All subsequent rounds were conducted in person on the college campus.

Round 1- (>1hr)

The round started with my introduction, followed by a few questions and cross questioning around RAG and LLMs, since I had mentioned an RAG project on my resume.

DSA - Question 1
Minimum time required to burn a binary tree

DSA - Question 2
I was given a list of names and a dictionary of valid names. The catch was that an OCR system removed the whitespace between words while scanning, so I had to determine which names could be reconstructed from the scanned text using the given dictionary. The interesting part was that the constraints and conditions were revealed gradually. I initially suggested a simple map-based approach. As more constraints were introduced, I moved to a set + greedy approach, and eventually arrived at a DP-based solution.

I was asked to write full code for both the problems on paper and explain the logic for time and space complexity. This round was less about knowing a particular problem and more about how I adapted my approach as the constraints changed.

Round 2- (1hr)

This round started with my quick introduction followed by questions on various topics.

Question 1 - Website Performance
Your hosted website is taking 10 seconds to load. What possible problems can you think of? I initially came up with around 5-6 possible issues, but the interviewer expected more. He guided me through the problem, and together we eventually identified around 9-10 possible causes. I was then asked to write all of them down.

We then went through my resume in detail, with quite a bit of cross-questioning on my projects. I put particular emphasis on my RAG project, which led to several follow-up questions.

Question 2 - Website Parsing Service
I was then asked to design the architecture of a website parsing service where websites could contain images, large PDFs and large amounts of text. I initially struggled to come up with a complete architecture, so the interviewer narrowed the problem down: How would you process a PDF of around 100 MB without losing context? I discussed different chunking and overlapping strategies, along with the possible use of LLMs.

Question 3 - Local Document Retrieval System
The final question was a case study. Suppose you are the manager of an event-organising company. You have a team of 6 people and an event deadline is tomorrow. The event is related to real estate, and you have 100 large PDFs containing potentially relevant information. You have a pendrive and need to distribute the PDFs to your team. The PDFs are too large to directly process using an LLM. The goal was essentially to build a local document retrieval system that could work under these constraints. I first ruled out using an LLM because of the constraints. I then considered dense retrieval, but the interviewer pointed out that we had no access to pretrained models and there wasn't enough time to train one. So I ruled that out as well. I eventually moved towards sparse retrieval and customised the approach according to the use case. I was then asked to write the Python implementation. I presented the pseudocode and overall approach, and the interviewer was satisfied with the solution.

Round 3- (45 mins)

The interviewer started by introducing himself. He mentioned that he had been with InfoEdge for over two decades. He asked me to introduce myself and then specifically asked me to talk about anything other than RAG and DSA, since those topics had already been discussed extensively.

I talked about my internships and focused primarily on my Trading Bot project. This led to several cross-questions about the project, along with questions such as: Why did you build a trading bot? Why not pursue finance? What did you learn from the project?

The interviewer then asked about my experience with AI models:

• Have you used any advanced/proprietary AI models?
• Which model do you like the most?
• Why do you prefer it?
• Is your preference based on benchmarks or actual experience?
• How would you justify that preference?

This was followed by questions around ML and Deep Learning, including:

• How does a neural network work?
• Design a neural network for a given problem
• What are activation functions?
• Name four activation functions
• Why are activation functions needed?
• How does an LLM work?
• What is the basic architecture of an LLM?

One of the more interesting parts of the interview was the discussion around my use of AI while developing projects. I mentioned that I use AI assistance regularly while coding.

• To what extent do you use AI?
• What do you use it for?
• Is it okay to rely on AI?
• Why build these projects yourself if an AI model can generate something similar with one prompt?
• How would you use AI in a professional environment?
• What factors would you consider before using AI?

The interviewer was essentially testing whether I understood the projects I had built and whether I could distinguish between using AI as a productivity tool and blindly depending on it.

The round ended with some generic questions: Why InfoEdge? Where do you see yourself in 5 years? What is your actual passion? What kind of work interests you?

Round 4- (20 mins)

This was a relatively short and generic HR round. I was told that this round was not intended to be eliminatory. Out of the 5 candidates who reached this round, 4 were selected. One candidate was rejected based on their performance in the previous rounds.

Final Result

Selected