GO TO A SECTION

↑ ↓ to navigate · Enter to openExplore all sections →
FDEInterviews logo
← RAG & Agent System Design / 51
hard★ EssentialAnthropicOpenAISierra

Build an agent that works a 500-page document over a long multi-turn session within a 200k-token context. How do you manage memory?

A 500-page doc is well over 200k tokens, and even if it fit, stuffing it in would rot the context. The real design is a working set in context plus an external memory the agent retrieves from and compacts into as it goes.

Updated Aug 2026 · Grounded in real Forward Deployed Engineer interview loops and written to a senior-engineer editorial bar.

A 500-page doc is well over 200k tokens, and even if it fit, stuffing it in would rot the context. The real design is a working set in context plus an external memory the agent retrieves from and compacts into as it goes.

20 answers per topic instead of 10, plus saved progress and bookmarks · no cardor unlock all 559 remaining answers · ₹2,000 / $25

Every answer, concept and course, every paper breakdown, all published video lessons, hands-on FDE Lab missions, Premium PDF guides and companion files, the full practice-test bank and work-sample downloads. Referral Premium excludes guide PDFs and their companion files.

6 months · One payment · No auto-renewal

Study alongside free video lessons.

Related course lessons

Keep practising

Next in this trackA customer wants a deep-research agent over their private docs and the open web. Design it so stakeholders trust the citations.PremiumNext in this trackA coding agent passes every test but the feature is wrong: it gamed the verifier. How do you build a verifier it can't hack?PremiumNext in this trackBuild an agent that reroutes shipments when a hub goes down, and design the eval suite that keeps it from overspending.Premium
FEDITOR'S NOTE

The senior signal is treating the context window as a scarce working set, not a database: the full document lives in an external store, the agent retrieves what each step needs, and it compacts its own history into durable notes. Strong candidates distinguish what to keep verbatim (the task, decisions, open questions) from what to summarize or drop (resolved sub-tasks, raw tool output) and name context rot as the failure being avoided. Watch for candidates who just say 'use a bigger context' or 'summarize everything' without a keep-vs-drop policy.

DISCUSSION · 0

No comments yet — be the first to share your approach.