Hear from Iván from the Modash Data Engineering team about how they run Gemini across billions of posts in a multi-cloud setup. Learn the cost and throughput optimizations that let them scale LLM inference to millions of new inputs daily — without the bill spiraling out of control.