Search results for incorrect api

🎯 Al/ML Engineer Interview — Scenario-Based Questions Round by Round If you're preparing for Al/ML, GenAl or Al Engineer roles, these are the companies where recent candidate reports show strong focus on RAG, Al Agents, LLMs, evaluation and system design Round 1 - Python + Problem Solving Your Python pipeline processes 10 million records and is extremely slow. How would you identify and fix the bottleneck? An API returns duplicate records randomly. How would you write a Python solution to deduplicate the data efficiently? Your ML preprocessing code works locally but crashes with a large production dataset. How would you debug it? You need to process a dataset that doesn't fit into memory. What approach would you use? Your inference API suddenly starts taking 3 seconds instead of 300 ms. How would you investigate the issue? Round 2 - ML + Statistics Your fraud model has 99% accuracy, but the business says it's missing too many fraudulent transactions. What would you change? Your model performs extremely well on training data but poorly in production. How would you identify the cause? Two models have similar accuracy, but one has significantly better recall. Which metrics would you consider before choosing one? Your model's performance drops after deployment. How would you determine whether it's data drift, concept drift or a pipeline issue? 5- You have very little labelled data. How would you build and evaluate an ML model? Round 3 - LLM + RAG + Al Agents 1- Your RAG application retrieves the correct document but generates an incorrect answer. How would you determine whether the problem is retrieval or generation? Your Al Agent repeatedly calls the same tool and enters a loop. How would you prevent and recover from it? Your RAG system works perfectly on 100 test questions but fails after deployment. How would you build a proper evaluation strategy? Your company wants an internal Al assistant using confidential documents. How would you design the RAG architecture and security controls? An Al Agent needs access to CRM, SQL database and external APls. How would you design the tool layer using MCP or APIs? Round 4 - Al/ML System Design Design a production-scale RAG system for millions of documents and thousands of concurrent users. Design an Al Agent that can research a topic, call multiple tools, verify information and generate a final report. Your LLM provider suddenly becomes unavailable. How would you design model failover? Your Al application has good quality but extremely high latency and inference cost. What would you optimize? Your Al system is deployed to production. What would you monitor? #AI #MachineLearning #GenAl #AlAgents