Staff Software Engineer

We are looking for an exceptional Staff Software Engineer to own the most visible surfaces in the company: the AI agent that programs industrial robots through plain conversation, and the vertical applications built around it. This role tackles one of the most ambiguous, high-leverage challenges in applied AI: turning a successful LLM-powered demo into a product customers rely on. Our verticals are the real jobs customers hire robots to do, like palletizing, welding, and machine tending. Each vertical is a product surface, with purpose-built features and user experiences, designed from the start for the agent to drive.

Your impact will be measured not just by the code you write, but by your ability to make the agent's quality measurable, harden it to release, turn our verticals into products customers love, and raise the technical bar of a small, founding-stage team working directly with our founder, product leadership, and first customers.

WHAT YOU'LL DO

🏗️ STRATEGIC ARCHITECTURE & LONG-TERM VISION

- Define the Future: Own the long-term product and technical vision for the AI agent and the vertical applications it powers, anticipating the scale, security, and product needs of LLM-powered robot programming a year or more out, including its core representation problem: how a language model safely reads, writes, and edits large structured automation programs.

- Design the Experience: Own how programming a robot through conversation should feel: the interactions, defaults, and guardrails that earn an operator's trust.

- Decouple and Scale: Design the agent's tool surface as clear, stable interfaces against our robot platform APIs, so that what the agent can do grows with the platform, not against it.

- Standard Setter: Set the engineering standards for a nondeterministic product: latency and token budgets as product requirements, and evaluation gates as the bar every release must clear.

🏎️ COMPLEX PROJECT LEADERSHIP & EXECUTION

- Product Ownership: Act as part-PM for your surface: work backwards from the product experience, spend real time with customers and operators, be opinionated on the roadmap, and own outcomes, not tickets.

- Build the Verticals: Ship vertical applications alongside the agent: purpose-built flows for palletizing, welding, and machine tending that encode how the work is really done, and that the agent can drive end-to-end.

- Drive the Program: Lead the agent's path from demo to release across our robot platform, QA, product, and AI teams: convert fast-moving asks into testable requirements and hold scope against a real ship date.

- Hands-on Delivery: Own production hardening end-to-end across the agent's TypeScript/Node stack: reliability, secure key management and usage controls, degraded/offline modes, and review- and QA-gated release engineering.

🛠️ PRODUCTION EXCELLENCE & ORGANIZATIONAL LEVERAGE

- Systemic Quality: Build the agent's evaluation discipline (datasets mined from real usage, regression gates in CI, model-graded scoring) so that every early-customer failure, in any vertical, becomes an eval case and every eval win becomes a release decision.

- Technical Multiplier: Mentor and coach the engineers around you, significantly contributing to their technical growth and to the quality of everything the team ships.

- Influence & Accountability: Bring rigor to how the company understands agent quality: shared metrics that leadership, product, and customers can trust.

WHO YOU ARE

We are looking for a product-minded, LLM-native engineer who has shipped AI to real users and gets excited about both the agent and the real-world automation tasks it will drive, like palletizing and welding. You want to understand how the work actually gets done on a factory floor, and you build for the people doing it. This is not a model-training research role, and no robotics background is required. Agentic coding tools should already be part of your default workflow.

SKILLS YOU'LL BRING

- Experience: 8+ years of professional software engineering, including hands-on AI work: shipping an LLM-powered product (agent, copilot, AI-native workflow), building model evaluation systems, or production data work for AI/ML. Ownership of an LLM product end-to-end is the strongest fit.

- LLM Product Engineering: Fluency in agent loops and tool calling, prompt and context management, structured output, model selection and fallbacks, and token/latency economics as engineering constraints.

- Evaluation & Quality: You've built eval datasets, harnesses, or regression gates for nondeterministic systems and used them to make real launch decisions.

- Product Thinking: You've owned what to build, not just how to build it: you talk directly to users, form sharp opinions about their problems, and put decisions in writing. Experience serving operational, domain-heavy users (manufacturing, logistics, field operations) is a strong plus.

- Design Sensibility: Strong opinions about use