AI Memory Interview Questions

**Checkpoint model** Each superstep persists channel values, metadata (thread_id, checkpoint_id), and pending writes. Enables resume after crash, human interrupt, or async long jobs.

Compress agent memory into durable distilled state while preserving goals, constraints, and tool outcomes.

Medium AI Agents interview question on stateful agents across sessions — architecture, trade-offs, eval, and production patterns.

Hard AI Agents interview question on agent memory: short-term, long-term, episodic — architecture, trade-offs, eval, and production patterns.

Hard scenario on agent memory — episodic vs semantic memory, vector stores, summarization, and model parametric limits.

Multi-agent orchestration is a senior/staff-level system design question gaining traction at Google DeepMind and Microsoft. Learn the supervisor pattern, shared state management, and how to avoid the 'too many agents' anti-pattern that sinks most candidate answers.

LangGraph has emerged as the go-to framework for stateful agent workflows, replacing brittle LangChain chains in production. Interviewers want graph thinking — nodes, edges, conditional routing, and checkpointing — not just API familiarity. Essential for any LangChain shop hiring GenAI engineers.

A production AI coding agent typically includes:

Prompt engineering focuses on instruction wording. **Context engineering** focuses on the full state fed to the model each step: policies, repository maps, retrieved files, prior tool outputs, memories, and task specs.

Medium Claude cost question on prompt caching breakpoints and hit rates.

Hard orchestration comparison — LangGraph agent graphs vs durable workflow engines.

**Session model** WebSocket/WebRTC streaming multimodal input/output.

**Storage** Vector + structured facts table keyed by user_id; version memories on conflict.

**Concept** Responses treat interaction as structured items (messages, tool calls, outputs) with native support for reasoning models and built-in tools — reducing glue code.

**Assistants API** Managed threads, code interpreter, file search, function tools with run lifecycle (queued/in_progress). Less boilerplate for CRUD-style agent apps. Trade-offs: vendor lock-in, run polling complexity, harder fine-grained graph control.

**Tracing** Span per node: inputs hash, duration, LLM/tool child spans. thread_id correlates user session.

**Discovery** Instrument legacy script to log state snapshots; draw FSM on whiteboard; identify hidden globals.

**Supervisor pattern** Central node decides next worker or FINISH based on plan and worker outputs — alternative to flat handoff messages.

**Unit layer** Pure functions: conditional edges, state merge logic, parsers — no graph compile needed.

**State schema** ticket_id, customer_tier, messages, retrieved_docs, proposed_actions, sentiment, escalation_reason, resolution_code.

**Retry patterns** Tenacity-style retries inside node or dedicated retry wrapper node incrementing state.retry_count. Exponential backoff for transient MCP errors.

**Parallelism model** LangGraph can schedule independent nodes in same superstep concurrently — e.g., search three sources at once. Reducers combine outputs into unified state field.

**Requirements** Durability across pod restarts, concurrent threads, query by thread_id, retention/compliance deletes, backup/restore.

**Stream modes** graph.stream() / astream_events yield incremental updates — node enter/exit, partial state, LLM tokens. Pick mode matching product needs: chat UI wants messages; ops dashboard wants updates.