Salman Munaf

Salman Munaf is a lead site reliability engineer at TikTok, working on video infrastructure and distributed systems. He previously worked as a software engineer at Meta.

He applies reliability engineering to agents that call external services. His approach uses idempotency, status checks, bounded retries and narrowly scoped credentials to control uncertain outcomes. He also treats agent memory as state that needs freshness and provenance, and human approval as permission for a specific action with a defined scope.