Nova Intelligence is building the most powerful AI platform for SAP teams. SAP is the heart of the enterprise: almost every critical business process at the world's largest companies — finance, supply chain, manufacturing, sales — runs on SAP. Our customers use Nova to 3x the productivity of their SAP teams: modernizing legacy code, redesigning business processes, resolving production issues, and more.
Nova Intelligence was founded by AI researchers, repeat entrepreneurs, and the co-inventor of SAP HANA, and has raised $40M from SAP.io, Accel, Conviction, and Chemistry.
About this role
We're hiring software engineers to build the platform that powers Nova. The role spans agent infrastructure, distributed systems, product engineering and more. The common thread is solving hard problems without off-the-shelf solutions. Enterprise systems like SAP weren't built for modern developer tooling, so we're building foundational infrastructure from scratch — the indexing, versioning, authentication, execution, and integration layers that agents need to understand, create, and transform code at scale.
We hire across senior, staff, and principal levels. We don't use formal levels internally — we look for people who can drive substantial work end-to-end — and we reflect seniority in compensation.
What you'll do
Build the core platform. You'll work on the systems that let the agent reliably navigate, understand, and operate on large, interconnected enterprise codebases — code search at scale, authentication, sandboxing, tool calling, and the agent orchestration that ties them together.
Make the agent reliable. You'll design evaluation frameworks and feedback loops that systematically improve agent performance on real customer workloads. You'll instrument what the agent does, surface failure modes, and turn customer scenarios into reproducible tests that drive measurable improvement.
Build for enterprise deployment. You'll ship the infrastructure that lets Nova run reliably at Fortune 500 scale — automated deployments across many customer AWS accounts, robust RBAC and permissioning for agents, security and compliance (SSO, audit, data residency), and cloud and on-prem environment support.
Ship product end-to-end. You'll build the things customers see and use daily — backend services, integrations with the rest of the enterprise stack, and the client applications customers live in every day.
What we're looking for
You're a strong software engineer with a track record of shipping production systems end-to-end. We work primarily in Python, JavaScript/React, and Go.
You have strong cloud and infrastructure chops. Familiarity with AWS is a must, and you're comfortable working in infrastructure-as-code (Terraform or similar).
You have informed opinions on agent design and frontier models. You've used Claude Code, Cursor, Cline, or similar enough to know what works and what doesn't, and you've optimized your own workflow to ship faster.
You're comfortable with ambiguity. You can take an ill-defined problem and ship a solution without waiting to be told how.
You communicate clearly across engineering and non-engineering audiences.
Background we'd love to see
3+ years of experience in backend, infrastructure and distributed systems engineering
3+ years of experience with cloud platforms (AWS, GCP), Kubernetes, or similar
Production experience with LLMs and AI agents
Time at an early-stage startup
Contributions to open-source projects in AI tooling, developer tools, or adjacent spaces
Why join
You'll work alongside top engineers, AI researchers, and SAP experts on technical problems few teams get to touch. What you ship runs inside the world's largest enterprises — applied AI on real, mission-critical systems with measurable economic impact.
We use Nova to build Nova. You'll be a power user of your own tools, obsessed with iteration speed and shipping fast.
We're a small, highly talent-dense team. We offer competitive compensation, meaningful equity, and full health, dental, and vision coverage.
If you read this and recognize the role you want, apply — even if you don't match every line.