Want to know if this job is worth applying to?
San Francisco, California, United States
Onsite
At Composio, we are building infrastructure that allows agents to communicate with the tools you use for work including Github, Gmail, Notion, Salesforce, etc. We are a small team of engineers wrangling problems from context to search, that help us provide the most capable bridge between your agents and your tools. We raised a $25M Series A from Lightspeed with angels like Guillermo Rauch (CEO of Vercel), Dharmesh Shah (CTO of Hubspot), and Gokul Rajaram. We've 3x'd our ARR this year, and our customers range from your friends in the YC batch to Wabi, Glean, Zoom and many more.
running stateful agent infrastructure inside someone else's cluster is new territory. this role defines what reliable, repeatable on-prem agent deployment looks like, across the customers running us in their own environments.
THE FRONTIER YOU'LL WORK ON
• fleet-managing agent runtimes across air-gapped, regulated, BYOK environments has no playbook. you define one.
• turning hand-managed on-prem deployments into a repeatable product. the difference between 5 self-hosted customers and 50.
• every incident, CVE, and upgrade wave here is unsolved at the ecosystem level. your runbooks lead the industry.
WHAT YOU'LL DO?
• own the on-prem lifecycle: helm charts, upgrades, the weekly release train, rollback paths.
• drive chart convergence: diverged customer forks back onto one supported chart, with guardrails.
• build structured onboarding (replicated or equivalent) so a new customer deploys in days, not weeks.
• debug across the stack: temporal workflows, k8s init ordering, storage backends, auth-refresh pipelines.
• own the security surface: CVE triage, patching, RCAs, bulletins. keep customers running when releases freeze.
• co-own per-customer tool patching with the platform team.
• dogfood Composio on your own deployment and support workflows.
"MUST HAVES"
if you are very good, nothing is a must per-se
• core infra engineering
– you've run software in environments you don't control: on-prem, VPC, air-gapped.
– deep kubernetes and helm: chart authoring, init-job ordering, multi-environment releases.
– you keep systems reliable while the product moves fast.
• fleet judgment
– you think in version skew, upgrade waves, and rollback paths across many instances.
– you separate one customer's bespoke ask from the primitive every install needs.
• ai native
– you have built with the language models.
– you have opinions on agents, MCP, and where they break in real deployments.
• linux - comfortable deep in the stack.
• typist - your RCAs and security notes earn customer trust.
• human - you build trust and admit what you don't know.
OPTIONAL
• temporal, Postgres, cloud secret managers (GCP / AWS).
• replicated or packaged self-hosted delivery (kots, airgap bundles).
• compliance: SOC 2, Vanta-style programs, security questionnaires.
• major open source contributions.
