Infrastructure Engineer

Nova is building an AI agent platform for SAP teams. SAP runs finance, procurement, supply chain, manufacturing, HR, and customer service for most of the Fortune 500 and the Global 2000. It is old, expensive, and slow to maintain, with its own programming language and object model, so tools such as Claude Code and Codex do not work on it. Enterprises spend roughly $200 billion a year on SAP services. Nova builds the platform that brings frontier models to the people who work in SAP, so those teams can modernize legacy code, redesign processes, and resolve production issues. Customers use it to multiply the productivity of their SAP teams.

Nova is about 30 people. It closed a Series A earlier this year and has raised about $40 million, including from SAP.io, Accel, Conviction, and Chemistry. The founding team pairs AI engineering with deep SAP experience. The CEO is a repeat founder whose prior company reached a few million users and stayed profitable. A co-founder is a former researcher at Google DeepMind and Facebook Research who previously started a company that raised about $10 million. The chief scientist co-invented SAP HANA and was global CTO for Accenture's SAP business, the largest SAP consulting practice. The Chief Customer Officer was CIO at Cargill, one of the most complex SAP environments in the world. Series A, about 11 to 50 people, founded in 2024, in New York.

Nova is hiring infrastructure engineers to keep building the platform. This is not a product seat. The domain is greenfield, with no ready-made answer. You build foundational infrastructure from scratch: indexing, versioning, authentication, execution, integration, code search at scale, sandboxing, tool calling, agent orchestration, and deployment across many customer AWS accounts, plus RBAC, SSO, audit, data residency, and support for both on-prem and cloud. These are critical enterprise systems that cannot break, so reliability, security, scalability, and observability are the bar.

The stack is Python, JavaScript and React, and Go, with AWS and Terraform at the center. The profile runs from a platform and distributed-systems engineer who has worked on large-scale systems to an infrastructure engineer who could write Terraform all day. AI experience helps and is not required. Zero-to-one experience is useful and not required. The work is about doing it correctly and answering customer questions with full confidence.

Development is extremely AI-driven. They want people who form opinions on the developer workflow and push developer experience forward. On a small team, moving developers from 3x to 4x more productive is where an engineer earns more than they cost. The first 30 days are exploration: solve issues, contribute to the codebase, learn the product, and work on internal tooling, which is lower-stakes than customer infrastructure. By about 60 days, a strong fit owns large pieces, such as code search or the deployment pipeline, and improves them rather than only maintaining them.

You own core platform and infrastructure: infrastructure as code, primarily on AWS, plus reliability, scalability, security, and the systems the agent relies on, including code search, authentication, sandboxing, tool calling, and orchestration. You own enterprise deployment across many customer AWS accounts, including RBAC, SSO, audit, and data residency. You own connectivity between Nova and customer environments: networking, VPNs, firewall rules, and routing, with customer IT and network teams. Where the seat leans security, you own the compliance posture, including ISO 27001 and SOC 2 Type 2, and you push the team's AI-native developer experience forward.

Bar: about 3 to 8 years. Strong platform, backend, or distributed-systems engineering, or deep infrastructure, DevOps, or security engineering. Hands-on Terraform in production. Strong AWS, including VPC, networking, VPNs, IAM, and compute, plus solid networking fundamentals. CI/CD such as GitHub Actions. Python, JavaScript, React, or Go. You can take a novel problem and work it out from first principles. Compliance familiarity with SOC 2 or ISO 27001 is a plus. They weight clock speed: a strong academic record, or competition results such as a high Codeforces rank. Time at a high-talent-density company in its strong period is the pedigree they name. The examples are Addepar in its earlier days, Ramp in any period, and Uber around 2016. Clear ownership of large infrastructure, and opinions on AI-native development, are green flags. About 7 to 10 years in big tech, about 10 years around startups with nothing that worked, someone who can only speak about familiar ground, or someone who needs a fully defined ticket is a miss.

In office five days a week in New York. Visa sponsorship is available. Equity is about 0.1% to 0.5%. Health, dental, and vision through Rippling. Posted work intensity is 9-9-5. The process is an intro with the CEO, two technical conversations with founding engineers, a paid work trial of 1 to 3 days, references, then an offer.