Member of Technical Staff, Release Engineering

Comfy•San Francisco, CA

7d•$180,000 - $220,000•Onsite

About The Position

Comfy ships multiple products across many repositories and services. Increasingly, our code isn't written by hand — it's written by agents, and our job is to keep up with what they produce. Our goal is continuous delivery: not removing humans from the loop, but removing the friction around them — so releases stop being a scheduled event and human judgment is spent on the decisions that matter, not on shepherding the pipeline. This is not an infrastructure role in the traditional sense. You won't own the production infrastructure our software runs on. You'll build the systems above it: the CI, cross-repo release orchestration, and ephemeral environments that let a fast-moving, AI-enabled team ship connected software with confidence. As code throughput accelerates, the delivery system becomes the bottleneck. Making sure it never is — that's your mandate.

Requirements

Have run Kubernetes and ArgoCD in production, and manage infrastructure through code.
Have built CI/CD for complex multi-service systems, providing a solution that multiple teams depend on.
Care about the reliability of software delivery systems, and have strong empathy for developers and internal teams who want to move fast on systems that just work.
Have strong opinions on the future of agentic coding — and have thought about the systems needed to handle the accelerated code throughput it produces.
Have a startup mindset: you thrive in fast-moving environments, take ownership, and love solving problems at scale.

Nice To Haves

Provisioned and managed GPU-based compute.
Worked on an open-source project.
Familiarity with ComfyUI.

Responsibilities

Build CI for an AI-enabled team. Build continuous integration systems that unlock development velocity for a team producing sharply accelerated code throughput.
Orchestrate releases across many repos. Design release processes that coordinate software delivery across multiple repositories in a multi-service backend, keeping our connected components continuously integrated.
Integrate release processes across teams. Bring multiple teams' release processes together so our connected software components ship in lockstep rather than drifting out of sync.
Give humans and agents somewhere to test. Provide ephemeral environments that spin up per PR, so people — and the agents writing our code — can test changes against a real, isolated deployment before anything merges.
Make quality part of the pipeline, not a meeting. Work with QA and development to inject test suites into release processes, so we ship quality at speed instead of trading one for the other.
Platform our creatives. Provide self-serve infrastructure for internal creatives to test and generate with cutting-edge models.