Agent / Applied AI Engineer

Rootstock Software

13d•Remote

About The Position

We are building the AI-first ERP, transforming the traditional ERP system into an AI-native experience. Our product surface allows users to describe their needs, and an agent fulfills them by surfacing the right actions, insights, and next steps. This is a fundamental shift from traditional ERP interfaces, offering a composable, AI-first experience that simplifies complexity. Rootstock has a strong foundation in ERP and manufacturing, with a clear vision and funding to become AI Native. You will collaborate with senior engineers experienced in shipping ERP at scale.

Requirements

Hands-on experience building with LLMs in production, including agents, tool-calling, MCP or equivalent, retrieval and grounding, and prompt/context engineering.
Shipped a real AI product and observed its behavior in the wild.
Strong engineering fundamentals and genuine product sense.
Competence in Java, JavaScript, or a comparable language.
A serious point of view on agent design, including scoping tools, grounding agents, safe failure mechanisms, and evaluation strategies.
Curiosity about the ERP domain; the ability to simplify complexity is valued over deep ERP knowledge.
Authorization to work in the US; visa sponsorship is not provided.

Nice To Haves

Experimenting with AI for fun, including building personal AI agents, bots, or tools.
Experience with rapid iteration, building proofs of concept, and debugging with evaluations.
Independent, self-directed, and motivated to make a significant impact.
A strong sense of taste and the ability to distinguish between genuine AI features and gimmicks.

Responsibilities

Design and build agents, tools, and AI-native interactions to transform a complex ERP into a user-centric experience.
Develop agents with authority over real workflows, capable of taking action within business processes like orders, inventory, and financials, ensuring safety and correctness.
Build typed, intent-routed tools over MCP (likely referring to a specific internal system or framework), addressing challenges in scale and performance for ERP's vast surface area.
Create a composable, AI-first UI using a typed component library that can be driven by both humans and models.
Develop and own the evaluation harness for measuring agent performance and ensuring continuous improvement of the user experience.