Senior Manager, AI Corporate Engineering

At Vanta, our mission is to help businesses earn and prove trust. We believe that security should be monitored and verified continuously, and we empower companies to practice better security and prove it with ease. Vanta has a kind and talented team, and while some have prior security experience, many have been successful at Vanta without it. Vanta's Corporate Engineering team is the infrastructure layer that keeps 1,500+ people connected, secure, and moving fast. AI is now central to how that happens, and this role leads the team that owns the backbone underneath it. Corporate Engineering owns the shared AI platform for Vanta's internal systems: identity and access for AI tools, cost visibility and controls, sanctioned tooling and guardrails, and the enablement that helps people use those tools well. We do not own product AI, which stays with Engineering, and we do not replace the AI work happening inside individual departments. We build the engineering layer that makes all of it safer, cheaper, and better supported. The company has moved fast. AI assistant use is widespread across every function, MCP infrastructure is live and self-serve company-wide, internal apps ship on a hosted platform, and AI spend has grown to the point where attribution and guardrails genuinely matter. What does not exist yet is a single owner for that platform layer. That is this role. As Sr. Manager, AI Engineering, you will lead a small, senior team, write the charter for what Corporate Engineering owns versus enables, and build the platform that lets the rest of Vanta adopt AI quickly without accumulating cost, risk, or duplication. What you’ll do as a Senior Manager, AI Corporate Engineering at Vanta: Lead, coach, and grow a senior team spanning platform engineering and technical program management. Set clear direction, hold a high bar, and give the team explicit permission to push back with data. Own the AI access layer end to end: identity and authentication for AI tools, connector and integration allowlisting, tiered permissions by function, and guardrails that make adoption safe by default rather than safe by exception. Own AI cost engineering. Build attribution down to team and individual, right-size model selection to actual need, make agent and automation spend traceable, and set budget thresholds and alerting in partnership with Finance. Own the MCP layer for internal tools end to end, including commissioning, decommissioning, governance, and the downstream impact assessment that has to happen before a migration ships rather than after it breaks. Own the paved path for internal applications: a golden-path deployment pattern, sensible defaults including private-by-default, lifecycle and decommissioning policy so unused apps retire themselves, and a build-versus-buy framework that keeps people from rebuilding tools we already own. Build the enablement layer that turns adoption into impact: a skills and agent registry that cuts duplication, an evaluation framework for internal AI usage, and technical enablement on sanctioned tools that goes beyond provisioning access. Write and hold the charter. Define what Corporate Engineering owns, what it enables, and how work enters the team, then convene the cross-functional forum that keeps departmental AI efforts coordinated rather than duplicated. Partner across Engineering, GRC, Security, Data, Finance, and People. Several of the levers this role depends on sit in those organizations, so the job is to coordinate and influence rather than override. How to be successful in this role: Experience standing up an internal AI or developer platform function rather than inheriting a mature one. You have written the charter, defined intake, and built the roadmap from a blank page. The ability to hold scope. You can take a commitment that was made without scoping or resourcing, land a bounded first version, and decline the rest without it reading as abdication. Fluency in AI cost engineering: usage attribution, model right-sizing, token economics, agent cost traceability, and the difference between a hard limit and a useful guardrail. An enablement-first instinct. Your default question is how to make something safe to self-serve. When you do have to decline access, you always explain why and offer an alternative path to the same visibility. Hands-on technical depth across LLM platform tooling, agent and MCP architecture, identity and OAuth, CI/CD, and infrastructure as code. Enough to review a design honestly and tell the difference between real risk and reported progress. Comfort building for an audience that is no longer only engineers. A growing share of the people shipping internal tools are not developers, and the platform has to work for them without lowering the security bar. Data governance fluency in a regulated environment. You know what should never enter a prompt, a repository, or a third-party tool, and you can make that judgment fast without becoming the bottleneck. Comfort