Started with “Tell me about yourself” and then went into detailed project discussion. Questions covered project architecture, workflow, technology choices, alternatives, LangChain vs LangGraph, FastAPI and its methods. They also discussed an MLOps project, including pipeline, limitations, failures and technical decisions. There were questions about recent AI developments and a behavioural question about working outside my area of expertise. The round lasted around 45–50 minutes.

Asked to design an end-to-end pipeline for processing millions of emails, focusing on distributed systems, databases, partitioning, sharding, indexing, query optimisation and large-scale data processing. Also asked to explain an LLM to a 6-year-old, followed by questions on LLMs, new data, limitations and AI concepts. Internship experience and a behavioural question about a time when my work was appreciated were also discussed. The round lasted around 35 minutes.

Tips:

Prepare projects thoroughly—know the architecture, workflow, technology choices, alternatives, limitations and failures. Also prepare DSA, DBMS/SQL, OOP, OS, Computer Networks, System Design, Design Patterns, SOLID and basic Distributed Systems. For AI/ML roles, know basic AI concepts and a few recent AI developments. The interviewer focused heavily on the candidate’s thought process rather than memorised answers.