Valuelabs

Group Discussion based on Vertical AI Consulting - Should discuss both gaps and solution. A lot of people were rejected in this round. Technical Interview 1: What is chunking Langchain libraries R2 score def What does regression work ig gradient descent they want to hear Rag pipelines in detail Linear regression loss function Tool calling in agents What are embeddeding models examples Explain your project DB Schema Technical Interview 2: Grilled on OOPS OOPs concepts: The interviewer asked if I was comfortable in Java or would I want to stick to C++, my original choice. I thought multi language familiarity would be a bonus point. In C++, you cannot instantiate an abstract class but in Java, you can write A obj=new B() or something like that when A is an abstract class and B extends it. I thought, if C++ gives some linked error and needs pointer referencing, Java should also give error. But that's not the case. The interviewer specifically told me after this to always stick to the language I know. Because interviewers care less about the tech stack and more about fundamentals unless it's a specific cluster or a specific role. DSA Question : Similar to Leetcode Qn : count and Say Rejected in round 2.