Senior Founding AI Engineer — Agent Runtime

As our Senior Founding AI Engineer - Agent Runtime, you will own the execution system behind an autonomous AI sales agent.

This is not a typical LLM application. You’ll be building a policy-driven execution system where model outputs are constrained, evaluated, and enforced by a deterministic runtime.

Reporting directly to the CTO/Co-Founder, you will play a critical role in shaping both the technical architecture and product direction. The systems you build will negotiate real contracts, protect real revenue, and operate within real-world constraints. The quality of your engineering is the difference between an AI that closes deals and one that gives away margin.

What You’ll Do

• Own the end-to-end execution system — including the agent pipeline and the event-driven runtime that governs lifecycle, state, and policy enforcement
• Design and evolve a multi-stage agent pipeline (intent classification, context assembly, reasoning, response generation) as a cohesive, testable system
• Build and maintain a robust evaluation framework — defining correctness and catching regressions before they reach customers
• Work within a structured rules and policy system — including constraints, escalation logic, and commercial guardrails
• Design systems that are safe by construction, ensuring the agent operates within pricing, legal, and approval boundaries at the architecture level
• Architect context assembly — determining what to include, retrieve, compress, or discard as complexity scales
• Build instrumentation and feedback loops so every interaction improves system performance over time
• Make the agent configurable and increasingly self-sufficient — start from the rules, guardrails, and communication guidelines customers define today; build the instrumentation and feedback loops that reduce how much explicit configuration is needed tomorrow
• Lay the foundation for a self-improving system, where outcomes drive better models, smarter context selection, and improved decision-making
• Engage with early customers to validate assumptions and translate real-world usage into product direction