Skip to main content
← engineering

Metorial Builds Serverless Runtime for MCP and Opens Remote Engineering Roles

By Zero G Talent•

Connecting AI models to external databases, enterprise APIs, and workspace tools remains a messy systems engineering problem. While the Model Context Protocol (MCP) established an open standard for tool calls and context sharing, operationalizing MCP servers requires credential management, per-user permission boundaries, OAuth rotations, container orchestration, and audit logging.

Metorial builds the infrastructure layer designed to handle that runtime burden. Describing itself as "Vercel for MCP," the San Francisco-based startup provides a serverless runtime that hibernates idle MCP servers and boots them with sub-second cold starts when a request lands. Backed by Y Combinator and Pioneer Fund, the four-person company is hiring engineers to scale its distributed systems and deployment infrastructure.

The Plumbing Beneath Production Agents

AI agents frequently stumble when moving from local prototypes to production environments. Writing a Dockerfile or configuring a local integration is straightforward, but deploying agentic workflows across hundreds of enterprise tenants introduces operational complexity. Developers face recurring hurdles: maintaining OAuth flows across distinct services, tracing tool calls across sessions, and isolating customer credentials.

Metorial abstracts this layer into a managed, multi-tenant execution environment. The platform includes a catalog of verified integrations spanning software-as-a-service applications, databases, and custom MCP servers. Through a feature called Magic MCP, developers can point an MCP-compatible client at a single URL to access pre-configured integrations. The company also offers client SDKs across TypeScript and Python to instantiate integrations in single function calls.

On the security side, the company built Protoguard to inspect tool requests and filter prompt-injection attempts before they execute on connected APIs. The runtime maps tool executions onto enterprise identity systems with role-based access control, SAML SSO, and audit logging for session forensics. In a demo on YouTube, the team demonstrated deploying three MCP servers—including Slack and Google Calendar with OAuth authentication—and connecting them to an agent workflow in roughly 30 lines of code.

Ticketing Roots and Systems Architecture

Metorial was founded in 2025 by Chief Executive Officer Karim (Wen) Rahme and Chief Technology Officer Tobias Herber. The two met at age 14 as seatmates in an Austrian technical high school, spending five years building projects together before entering production software development.

Prior to Metorial, Rahme and Herber co-led technical development at Valicit, an Abu Dhabi-based event ticketing platform. As reported by yespress.io, Valicit handled ticketing for high-volume concerts featuring Travis Scott, 50 Cent, and G-Eazy, ultimately processing over 2 million tickets while fielding drops with up to 50,000 concurrent purchasers. Handling spiky, unforgiving ticket demand shaped their approach to system reliability and distributed state.

Rahme later studied computer science and mathematics at New York University Abu Dhabi, while Herber completed bachelor's and master's degrees in computer science at Johannes Kepler University of Linz, earning an IEEE best paper award in systems software. After initially building a no-code AI workflow tool, the founders pivoted to address the operational glue they kept re-implementing. Following an initial rejection for YC's Summer 2025 batch, the team grew open-source adoption, signed early commercial pilots, and secured entry into YC's Fall 2025 cohort.

Growth, EU Expansion, and Serverless Execution

Metorial operates on an open-core model. The core platform is open source under an Apache-2.0 license and self-hostable, while the commercial product provides a managed serverless runtime billed per request, alongside enterprise compliance and governance features. According to reporting from yespress.io, early users include engineers at FAANG companies, Big Four firms, and American Express.

During 2026, the company shipped version two of its platform and established an EU data-residency region to accommodate European compliance requirements. While MCP provides the current technical interface, Rahme has noted that the company's core proposition rests on managing permissions, tokens, and audit trails rather than depending on any single AI protocol.

To expand its platform capabilities, Metorial is hiring for three technical roles across distributed systems and cloud operations. The team is recruiting a Founding Engineer to develop distributed systems using TypeScript, Go, and optionally Rust, integrating AI agents via MCP while managing infrastructure on AWS and HashiStack. The company is also seeking a Backend Engineer to construct backend services and tool-calling functionality, alongside an Infrastructure/DevOps Engineer to oversee large-scale production operations on AWS and HashiStack.

All three positions are fully remote and require strong fluency in TypeScript, Go, distributed systems, and modern AI agent protocols.

Open Engineering Roles

Role Location Work Mode Compensation
Founding Engineer Remote Remote $100,000 – $180,000
Infrastructure/DevOps Engineer Remote Remote $100,000 – $170,000
Backend Engineer Remote Remote $100,000 – $165,000

See all open roles at Metorial