Concept guide · LangChain, LangGraph, and CrewAI
How do you keep a framework integration maintainable?
A slower explanation for when the short interview answer is not enough. Start with the intuition, then connect it to the technical idea and the way an interviewer expects you to reason about it.
ML area
Generative AI → Framework maintenance
Interview format
Advanced · Defense
Question
GENAI-100
Start with the idea
In simple terms
Keep framework code behind small interfaces, pin compatible versions, record prompts and state schemas, and test upgrades against real traces. Fast-moving frameworks are dependencies to manage, not foundations to hide everywhere.
Make it concrete
The application calls its own `runResearchWorkflow` interface, while the implementation can use LangGraph or CrewAI without exposing framework-specific objects to every feature.
Build the answer
How to defend your decision
- 1Define stable application interfaces around framework-specific code.
- 2Pin, test, and observe upgrades before broad rollout.
- 3Keep migration, rollback, and provider fallback paths explicit.
Go one level deeper
The technical explanation
Framework changes can affect message formats, retries, tool binding, state persistence, streaming, and default model behaviour. Isolate adapters, maintain migration fixtures, use release notes and upgrade tests, keep raw provider visibility, and maintain a direct fallback for critical operations where practical. The goal is reversible architecture, not avoiding all dependencies.
Say this in the interview
Pin compatible versions, isolate framework code behind small application interfaces, record prompts and state schemas, avoid leaking framework objects through the whole codebase, test upgrades against traces and evaluations, and keep a direct fallback for critical provider calls where practical.
Keep preparing