Dominik Tornow is the founder and chief executive of Resonate HQ and author of Think Distributed Systems. He builds programming models and infrastructure that help distributed applications—and increasingly AI agents—survive crashes, unreliable networks, and long-running operations without losing their progress.
Tornow studied at the Hasso Plattner Institute before holding engineering leadership roles at SAP and working as a principal engineer at Cisco and Temporal. At SAP, he focused on development tools, engineering processes, and collaboration; at Cisco, his interests included Kubernetes networking and service meshes. His subsequent work at Temporal sharpened his conviction that durable workflows should shield programmers from distributed-system failures much as database transactions abstract away difficult consistency problems.
At Resonate, he developed Distributed Async Await, a programming model that extends familiar asynchronous code across processes and machine failures. The company’s open-source projects include a durable-execution server, SDKs spanning TypeScript, Python, Rust, Go, and Java, and a Lean 4 protocol specification. His writing on durable execution defines interruption-tolerant computation independently of any particular framework, while Systems Engineering for Agentic Applications applies distributed-systems principles to autonomous software.
Durable promises and tasks. Tornow organizes recovery and coordination around a compact protocol built from durable promises and durable tasks. Persisting execution state allows applications to resume after interruptions, including agent workflows delayed by external services, human approvals, or expensive model calls.
The specification is the product. Tornow argues that coding agents will shift reusable infrastructure value from general-purpose implementations toward abstract protocols. The same specification can yield implementations tailored to Postgres schemas and transactions or NATS queues, versioned key-value storage, and scheduled messages. His AI Engineer conference talk shows why production reliability requires an intermediate, infrastructure-specific specification: code generated directly from an abstract protocol may pass basic tests while failing under concurrency or network disruption.
Deterministic simulation testing. Tornow builds reproducible simulation into Resonate to expose stale reads, optimistic-concurrency conflicts, and failing operation sequences. Simulators can reveal diagnostic state unavailable to production algorithms, helping coding agents discover correct distributed designs before generating concrete specifications and implementations.
Generating infrastructure from a specification requires more than passing basic tests: agents need a way to explore distributed algorithms, reproduce failures, and turn tested designs into concrete implementations.
What happens when the implementation becomes unnecessary?
