Loading Question...
In the technical round, the interviewer went through my resume for 3 minutes and asked me about some points in it, and he also asked follow-up questions. After that he asked me what topics i had prepared for the interview. I answered OS, OOPS, and DSA. I heard before my interview that different people are being asked questions on different topics. I was asked two questions each on OS and one on OOPS.
First one was to explain about mutexes, and the follow up was to explain how I'd use mutexes to limit number of processes accessing my database (mentioned in my project). Third question was to explain what happens in each case of the following when a class A and a child class B are defined with two functions with the same name in them:
The HM round began with my introduction and in the same intro, I also mentioned about my projects in the resume. He asked me to explain about each of the projects one after the other, what the project does, why the project was done, and the technologies and other technical things involved in it. This took nearly half an hour with both to and fro conversations.
While I was explaining about each of the projects, he brought up hypothetical situations regarding the projects and asked me what I'd do, why I'd do them and how. For example, one line in my resume mentions about the project being evaluated in a single threaded environment. He asked me if I would still keep it single threaded or multi thread safe if I were to deploy it for production. The follow-up questions went through explaining how multi threading can be achieved, what I'd do if each of the requests from user is processed by different threads, and how a database handles multi threading or multiple users parallelly.
Last updated Oct 19, 2026
After I completed explaining my last project, he brought up a hypothetical situation. One of my projects is about an RL being used to manage portfolio. He mentioned that if large number of people were using my model and due to some false rumors, the stocks took a wrong dip and he model sold some stocks, the selling of stocks would further decrease the price, and model would further sell, and this loop continues. He said that although my project aim was not to consider this, but these types of things are good for discussions.
Later I asked him what happens in some of the hypothetical situations. Like, in the future, after a point, almost all of the codebases will be handled by AI agents. Previously these agents were trained on large data, codebases coded by humans. But later they would be training on their own creations. I asked if there would be any actual development through their learning(training) or will the curve be a plateau. There was some 10 min discussion on this. He said the agents will need external, uninfluenced data to be trained on along with the codebases. Like comments from a code-reviewer.
After that I asked him another question. Say, a code reviewer is an AI agent, and all the repos it reviews are also managed by different AI agents. The code managers train on their own codebases and comments from reviewer, and reviewer trains on all these codebases. I asked if there will be any point of time after which we can swap any two agents A&B between two repos C&D, and all the other things like prompts, codebase, etc are not disturbed, and still get the same response from A at D as B gave at D previously and vice versa. This discussion continued for around 15 minutes or so.
At the end I asked him about himself, and the interview concluded.