Blog

Thoughts on AI, system architecture, and building scalable products.

2026-07-26Infrastructure

Processing 2,000 Hours of Media a Day with Celery and Redis

What running a video dubbing pipeline at 2K hours of daily content taught me about queue design, worker acknowledgment, and where a 50% speedup actually comes from.

11 min readCelery, Redis, Distributed Systems

Read note

2025-01-15AI/LLM

Scaling LLM Applications: Context, Queues, and the Cost of Personalization

What 100K+ daily requests and AI workflows for 1,000+ businesses taught me about context selection, asynchronous execution, cost, and quality.

9 min readLLM, Production, Architecture

Read note