Jordan Dziat
Senior Staff Engineer. Builder of systems that scale.
15 years in infrastructure and engineering leadership. Currently at The Lifetime Value Co., building platforms and tooling for engineering teams. I write about infrastructure, AI systems, and the craft of building software.
Notes
go test -race ./... doesn’t catch data races in code that isn’t exercised by your tests. If you have concurrent code paths that only trigger under load, you need to write tests that actually exercise them concurrently. Sounds obvious in hindsight.
Astro 5 shipped with content layer, server islands, and simplified prerendering. The content layer API is a massive improvement for managing blog content: glob loaders replaced the old file-based routing, and the Zod schemas catch frontmatter errors at build time.
The functional options pattern in Go is one of those rare cases where the community has converged on a genuinely good API design pattern. It’s more boilerplate upfront but the resulting APIs are impossible to misuse. Worth the investment for any public-facing library.
Latest Writing
A Team of One and a Hundred Agents
The unit of engineering output is no longer the person. It's the person plus the harness they've built to direct a fleet of agents, and the leverage is real, right up to the ceiling where judgment refuses to scale.
7 min readTaste Doesn't Come in the Diff
Agents optimize what they can verify: does it render, do the values map, do the tests pass. They cannot optimize what they can't measure, whether the result is legible, calm, honest, good. That gap is where the human reviewer's job moved.
8 min readVerifiable, Not Just Done
AI-led migrations are a genuine power tool. They sweep wide, mechanical changes across a codebase at a speed a solo developer can't touch. They also migrate the happy path and leave the rest to rot quietly, unless you design the migration to prove itself rather than just finish.
9 min readProjects
Nocturnium
An autonomous AI development platform where agents plan, code, test, and iterate on software overnight and open a production-ready pull request by morning. Built on the Hivemind multi-agent engine (agent orchestration, tool execution, persistent context and memory, and an LLM load balancer), running on the llm-go-sdk and simple-durable-jobs primitives.
Stout
A universal package registry that builds from source with integrated supply chain security. Supports Go, npm, Helm, Docker/OCI, Ruby, Python, and AI agent packages. Sandboxed builds, vulnerability scanning, SBOM generation, Sigstore signing, and a community trust system with reviewer verification.
Simple Durable Jobs
A lightweight Go library for building resilient, durable job queues with checkpointed multi-step workflows that automatically recover from failures. Includes fan-out/fan-in parallelism, priority queues, cron scheduling, and an embedded monitoring dashboard. Runs in production as the durable-execution engine behind Stout and the Hivemind agent platform.