Agent Harness Engineer

This role is with one of Dex's trusted partner companies. We work closely with their teams to truly understand their culture, goals, and what they're looking for, so we can match you with the right opportunity and give you context about the role before you commit to a process.
If you're interested sign up to Dex to apply.
Dex is an AI recruiter agent that helps you run your job search. Tell Dex your stack, seniority, and what you want to build. We will manage your applications and surface other opportunities that are a fit.
The role
This company is building a single, general AI agent designed to automate business tasks across thousands of integrations, processing over 1.5 million tool calls daily against real customer data. They believe a truly fluent agent, capable of operating across an entire company's tech stack, will fundamentally outperform a collection of narrow tools. Backed by leading investors and trusted by tens of thousands of teams, their mission is to bridge the gap between impressive AI demos and agents that reliably perform critical work.
You will build the core agent runtime, the loop where it plans, acts, and rigorously verifies its own work. This isn't about shipping features that might work; it's about making the agent measurably more fluent and reliable, growing its capabilities from trusted memory to self-writing skills. You will work directly with the founders, owning the system that turns every agent failure into a fix, with direct impact on user retention.
The work
Architect and build the core agent runtime, focusing on the planning, action, and self-verification loop.
Develop advanced agent fluency: tool composition across thousands of integrations, contextual recall, and dynamic skill management.
Expand the agent's autonomous capabilities, from trusted memory systems to self-writing skills and independent task execution.
Design and implement rigorous evaluation frameworks and regression tests to ensure measurable improvements and catch failures before they reach users.
Ship production-ready features daily, directly impacting user experience and retention across tens of thousands of teams.
What You Bring
You have personally built or significantly modified an AI agent harness, including its runtime, tools, memory, and evaluation systems. This is not a learning role.
You have a track record of making agents reliably close the loop, implementing verification mechanisms that prevent hallucinated success.
You think in systems, designing for hard technical trade-offs and anticipating failure modes at the scale of millions of daily tool calls.
You possess full-stack range, capable of moving from agent runtime to UI components to deployment scripts, and you ship to production daily.
You have read the foundational papers on agentic systems and hold strong opinions on evaluation methodologies.
Why apply through Dex
This is a rare, high-impact agent engineering role that won't be found on public job boards. Apply through Dex to cut through the noise, get a full brief on the company and role, and skip the cold application process. We connect you directly to the hiring team for roles like this, ensuring you're properly positioned from day one.
If you're interested, sign up to Dex to apply - https://meetdex.ai/job-listing?job_id=25cb5d89-701c-4b74-84cb-5a8f994a16f6&utm_source=linkedin&utm_medium=job_listing&utm_campaign=xml-feed
As part of the recruitment process at Dex, we process your personal data in accordance with our Privacy Notice for Job Applicants. This notice explains how and why your data is collected and used, and how you can contact us if you have any concerns.

Work arrangement
Yes