Async code vs background jobs

Must the HTTP response body include this work's result?
 YES → do it in the request path (async I/O), or
 enqueue + return 202 and let the client poll job status
 NO → Would users/business care if this work is lost on deploy?
 NO → BackgroundTasks is acceptable (keep it tiny)
 YES → Durable queue (command on RabbitMQ) or event (Kafka)

# Prefer async for concurrent I/O (DB, HTTP clients)
@app.get("/users/{user_id}")
async def get_user(user_id: int, db: Db = Depends(get_db)):
 return await db.fetch_user(user_id)

# Sync is OK if the whole stack is sync and you understand the thread pool
@app.get("/report-sync")
def report_sync():
 return heavy_cpu_report() # better: offload to worker

# PSEUDOCODE : OK: best-effort
@app.post("/events")
async def track(event: EventIn, bg: BackgroundTasks):
 bg.add_task(log_event_best_effort, event)
 return {"ok": True}

# PSEUDOCODE
@app.get("/users/{user_id}")
async def get_user(user_id: int, db: Db = Depends(get_db)):
 user = await db.fetch_user(user_id)
 if not user:
 raise HTTPException(404)
 return user

# PSEUDOCODE
@app.post("/exports", status_code=202)
async def start_export(body: ExportIn, db: Db, bus: JobBus, user=Depends(auth)):
 async with db.transaction():
 job = await db.insert_job(
 type="export_csv",
 status="pending",
 entity_id=str(user.id),
 idempotency_key=body.idempotency_key,
 payload_ref=body.filters_ref,
 )
 await bus.publish(
 queue="jobs.heavy",
 message={"job_id": str(job.id), "type": "export_csv"},
 )
 return {"job_id": job.id, "status": "pending"}

# PSEUDOCODE : DO NOT DO THIS for mail users expect
@app.post("/signup")
async def signup(body: SignupIn, bg: BackgroundTasks):
 user = await create_user(body)
 bg.add_task(send_welcome_email, user.id) # lost on deploy/crash
 return user