what's your strategy for handling background jobs in a microservices architecture?

hey everyone, ive been thinking about how to best manage background tasks like sending emails, processing images, or generating reports in a microservices setup. Are you typically using dedicated job queues (like RabbitMQ, Kafka, or SQS), or do you have services that poll for work? What are the trade-offs youve seen with different approaches, especially concerning reliability and scalability? Looking forward to hearing your experiences!