Chains, prompts, message types, output parsers, and the LCEL pipe syntax. Tests whether you can build and debug real LangChain pipelines — not just recite the docs.

RAG pipelines, chunking strategy, retriever patterns, and tool-based agents. Covers the decisions that separate functional prototypes from production-grade retrieval systems.

State machines, reducers, conditional edges, checkpointing, and interrupt-resume patterns. Tests whether you truly understand LangGraph's execution model — not just its API surface.

LangSmith tracing, evaluation design, and when to pick LangChain vs LlamaIndex vs raw SDK. Tests practical judgment on tooling decisions, not just feature lists.

Simulates a real entry-level screening round. One question per major framework area. Tests clean mental models, API contracts, and the traps that trip up developers in their first few months with LangChain and LangGraph.

Second easy mock round with fresh scenarios. Covers type contracts, retriever API changes, tool docstrings, and state machine basics. All questions target the same depth — different angles from Mock 1.

Simulates a real mid-level technical interview round. Applied reasoning, debugging production failures, and architectural design choices. Broad topic coverage with deliberate traps in every question.

Second medium mock round — fresh questions, different failure modes. Tests streaming internals, retriever debugging, agent concurrency bugs, and LangSmith evaluation design. Assumes comfort with the basics.

FAANG-style hard interview simulation. Edge cases, production gotchas, and non-obvious architectural trade-offs across all 8 topics. Expect multi-step reasoning, security traps, and questions where every option sounds plausible.

Second hard mock interview — fresh hard questions covering LCEL concurrency bugs, RAG multi-chunk synthesis, agent memory leaks, checkpointer race conditions, and LLM judge bias. No overlaps with Mock Hard 1.

Staff/senior engineer screening round. 18 questions targeting production failure modes, memory leak diagnosis, concurrency bugs, security vulnerabilities, and multi-step tradeoff reasoning. Designed to expose the gap between engineers who have read the docs and engineers who have debugged production systems.

AI architect and principal engineer screening. 18 questions on multi-agent architecture decisions, distributed state management, cost optimization at scale, evaluation rigour, CI/CD regression detection, and framework migration strategy. Every question requires reasoning across multiple systems simultaneously.