Comparison questionMedium~10 minAI Application Architecture / 7 of 18

Interview question

When should an AI task run synchronously or in the background?

Chooses request-response, streaming, or durable background execution based on duration, user interaction, reliability, and side effects.

1Say itAnswer out loud first

2CompareCompare with Core answer

3RefineRefine with Stronger and Advanced

TL;DR

I keep short interactive tasks synchronous when the user needs an immediate answer and the work fits a bounded timeout. I use background processing for long generation, large files, batch work, retries across dependencies, or tasks that must survive client disconnects. The API returns an operation identifier and exposes status, result, failure, and cancellation explicitly.

Full answer below

Practice focus

Execution model, durable work, cancellation, status, and user experience.

Answer first

Answer path

Core answer first, then checkpoints toward a stronger answer.

Start
Core answer
Start here. This is the direct spoken answer to practice first.
Overview
A task that may outlive the client request needs a durable execution contract.
I keep short interactive tasks synchronous when the user needs an immediate answer and the work fits a bounded timeout. I use background processing for long generation, large files, batch work, retries across dependencies, or tasks that must survive client disconnects. The API returns an operation identifier and exposes status, result, failure, and cancellation explicitly.
Checkpoint
Stronger answer
Optional upgrade with examples, trade-offs, and sharper phrasing.
Streaming can improve perceived latency for interactive text, but it does not make a long task durable. A queued worker claims work idempotently, renews or completes its lease, stores progress, and publishes a terminal result. The client can poll, receive a webhook, or subscribe to updates without coupling job lifetime to one web server.
Finish
Advanced answer
Optional deeper angle for production judgment and nuance.
I define admission limits, per-tenant quotas, deadlines, priority, poison handling, and retention before accepting background work. Cancellation is cooperative and distinguishes stopping future computation from reversing a completed side effect. Duplicate delivery is expected, so steps use stable operation keys and persisted checkpoints. Sensitive inputs are encrypted and expire on schedule, and provider-side background state follows the same deletion and residency policy.

Review signals

Use these to catch weak phrasing, shallow reasoning, and risky claims.

Common mistakesCandidateCommon mistakesMistakes that weaken the answer or signal shallow understanding.Open details

Details

Keeping an HTTP request open for an unbounded job.
Assuming streaming survives a server restart.
Treating queue delivery as exactly once.

Red flagsCandidateRed flagsSignals that an interviewer may treat as risky.Open details

Details

A client retry creates a second expensive job.
Jobs have no terminal failed or cancelled state.
The worker can process without tenant admission limits.

Follow-up prompts

Quick interviewer prompts inside this drill.

Follow-up questionsCandidateFollow-up questionsLikely next questions after the main answer.Open details

Details

Treat these as the next interviewer turn. Answer one out loud, then save the gap in your practice notes.

When is streaming enough?
How would a client resume status checks?
What does cancellation guarantee?

Continue the section

Next in AI Application Architecture: Design AI response streaming

Step 8 of 18

How would you design streaming responses for an AI feature?

Practice route

Read core answer Review checklist Follow-up prompts Track progress

Finished this drill?

Practice freely. Sign in only when you want to save completion, confidence, notes, or bookmarks.

New to Aporeon? Create account

Drill details

Human editorial direction

Aporeon is shaped by Aleksandar Tomovski, a software developer with experience on both sides of technical interviews. Content is reviewed for accuracy, natural spoken delivery, useful depth, and honest trade-offs.

About the creator Editorial standards

Was this drill useful?

Help improve the interview library.

Found this answer useful?

Say thanks with a standalone, one-time $5 contribution. No account required.

No subscription or perks.Choose another amount

Opening drill

Comparison questionMedium~10 minAI Application Architecture / 7 of 18

Interview question

When should an AI task run synchronously or in the background?

Chooses request-response, streaming, or durable background execution based on duration, user interaction, reliability, and side effects.

1Say itAnswer out loud first

2CompareCompare with Core answer

3RefineRefine with Stronger and Advanced

TL;DR

Full answer below

Practice focus

Execution model, durable work, cancellation, status, and user experience.

Answer first

Answer path

Core answer first, then checkpoints toward a stronger answer.

Start
Core answer
Start here. This is the direct spoken answer to practice first.
Overview
A task that may outlive the client request needs a durable execution contract.
I keep short interactive tasks synchronous when the user needs an immediate answer and the work fits a bounded timeout. I use background processing for long generation, large files, batch work, retries across dependencies, or tasks that must survive client disconnects. The API returns an operation identifier and exposes status, result, failure, and cancellation explicitly.
Checkpoint
Stronger answer
Optional upgrade with examples, trade-offs, and sharper phrasing.
Streaming can improve perceived latency for interactive text, but it does not make a long task durable. A queued worker claims work idempotently, renews or completes its lease, stores progress, and publishes a terminal result. The client can poll, receive a webhook, or subscribe to updates without coupling job lifetime to one web server.
Finish
Advanced answer
Optional deeper angle for production judgment and nuance.
I define admission limits, per-tenant quotas, deadlines, priority, poison handling, and retention before accepting background work. Cancellation is cooperative and distinguishes stopping future computation from reversing a completed side effect. Duplicate delivery is expected, so steps use stable operation keys and persisted checkpoints. Sensitive inputs are encrypted and expire on schedule, and provider-side background state follows the same deletion and residency policy.

Review signals

Use these to catch weak phrasing, shallow reasoning, and risky claims.

Common mistakesCandidateCommon mistakesMistakes that weaken the answer or signal shallow understanding.Open details

Details

Keeping an HTTP request open for an unbounded job.
Assuming streaming survives a server restart.
Treating queue delivery as exactly once.

Red flagsCandidateRed flagsSignals that an interviewer may treat as risky.Open details

Details

A client retry creates a second expensive job.
Jobs have no terminal failed or cancelled state.
The worker can process without tenant admission limits.

Follow-up prompts

Quick interviewer prompts inside this drill.

Follow-up questionsCandidateFollow-up questionsLikely next questions after the main answer.Open details

Details

Treat these as the next interviewer turn. Answer one out loud, then save the gap in your practice notes.

When is streaming enough?
How would a client resume status checks?
What does cancellation guarantee?

Continue the section

Next in AI Application Architecture: Design AI response streaming

Step 8 of 18

How would you design streaming responses for an AI feature?

When should an AI task run synchronously or in the background?

Practice focus

Answer path

Overview

Review signals

Follow-up prompts

Continue the section

When should an AI task run synchronously or in the background?

Practice focus

Answer path

Overview

Review signals

Follow-up prompts

Continue the section