What we're working on
We build AI systems for complex enterprise environments where AI is expected to operate correctly inside production systems. The goal is not just to generate answers, but to safely read, diagnose, and act inside environments that are deeply connected, stateful, and high-risk.
The challenge is not generating output. It is making sure the system behaves correctly when that output is used.
MCP is central to how we build. We use MCP-based tools and agent workflows to connect AI to real systems in a controlled way.
You would own the backend systems that define how AI interacts with production environments.
Who you'd work with
A small senior team. You would work closely with the founder on technical direction and with the rest of the engineering team on shipping. Direct access, no layers, decisions happen in the conversation you are part of.
What you'd do in your first 90 days
Day to day, you will write Node.js and TypeScript, design APIs, work on MCP-based tools, execution pipelines, agent workflows, and system integrations, and define the control boundaries between AI and production systems.
You will spend a lot of time thinking about failure modes, edge cases, testing, regression validation, and what the system should do when something is unclear or wrong.
- Weeks 1–2: Learn the codebase, the product, and the MCP and integration patterns we use. You do not need domain experience coming in. You will pick up what matters quickly.
- Month 1: Ship your first feature end to end. Likely a new capability for the AI to safely read, inspect, or analyze something inside a complex enterprise system.
- Months 2–3: Own a significant part of the backend. Set direction on architecture questions. Start shaping how we handle reliability, performance, validation, and approval flows between AI and production systems.
What you need to be good at
- Senior backend engineering, with production systems you can point to
- Deep fluency in Node.js and TypeScript, or comparable depth in another stack with a credible plan to ramp quickly
- Designing systems that stay reliable under real-world load and imperfect data
- Strong judgment around failure handling, edge cases, and system correctness
- Working without a spec. You can define scope, make calls, and move forward
- Using AI coding tools such as Claude Code, Cursor, and GitHub Copilot with judgment on when to trust the output
Experience with MCP, agent workflows, developer platforms, or adjacent patterns is a plus. SAP or enterprise ERP experience is great to have, not required.
What you'd pick up here
- How enterprise systems actually behave under the hood
- Building LLM-integrated systems for enterprise use, with real audit, safety, and control constraints
- How MCP-based tools and agent workflows behave in production environments
- How enterprise buyers evaluate and adopt software, and how that shapes what you build
The deal
Competitive market rate. Milestone bonuses. Real ownership of your domain.
How to apply
Email contact@innvenza.com with:
- Which role you are applying for
- One or two production systems you have built, with enough detail to show how you work
- GitHub, resume, or both
- A paragraph on what draws you to this specifically
Bonus: A Claude Code or Cursor transcript from a session you are proud of.
Short, specific messages preferred.