QA Engineer (Agents)

Job description

ABOUT SANA Sana is an AI lab building superintelligence for work. We believe organizations can accomplish their missions faster when teams can effortlessly access knowledge, automate repetitive work, and learn anything with the help of agentic AI. As part of Workday, we are committed to building AI that augments people - not replaces them. We bring this mission to life through two products. Sana Agents provide a seamless way to access all your company’s apps, knowledge, and data, enabling AI agents to do real work so teams can process and act on information at unprecedented scale. Sana Learn is an AI-powered learning hub that combines the simplicity of a modern learning platform with intelligent features like an AI tutor, automated content generation, and interactive apps, making knowledge not just accessible but actionable. We’re a talent-dense, product-obsessed team of engineers and designers from companies like Google, Spotify, Apple, and Databricks, united by deep technical excellence and rapid iteration. Our tools already help over a million people learn and work better across hundreds of leading enterprises - and we’re just getting started. ABOUT THE ROLE You’ll be the quality champion for Sana’s AI agent platform, ensuring our LLM-powered products are robust, reliable, and a delight to use. You’ll design and implement test strategies that keep pace with rapid iteration, automate critical workflows, and drive a culture of quality across engineering. This is a hands-on role for someone who thrives on constructing scalable ways of breaking things, uncovering edge cases unique to agentic and LLM-based systems, and building the safeguards that prevent issues from reaching production. You’ll help us deliver agent workflows that are safe, trustworthy, and enterprise-ready, for the AI landscape of today and tomorrow. IN THIS ROLE, YOU WILL - Design and implement test plans for agent infrastructure, LLM-based APIs, and end-to-end user journeys - Build and maintain automated test suites for backend, frontend, and integration layers, including prompt and response validation for generative models - Develop tools and frameworks to accelerate testing and catch regressions early, especially in agent reasoning, tool use, and context handling - Collaborate closely with engineers to embed quality into every stage of development, with a focus on the unique challenges of AI/LLM systems (e.g., non-determinism, hallucinations, safety) - Lead root cause analysis and drive resolution for critical issues and incidents, including those arising from model updates or agent behaviors - Advocate for best practices in code quality, observability, and CI/CD pipelines—ensuring quality signals are actionable and visible WHAT SUCCESS LOOKS LIKE - Critical bugs, regressions, and model failures are caught before they reach users—even as we scale and ship rapidly - Automated test coverage is high, reliable, and easy to maintain, including for LLM outputs and agent workflows - Release cycles are fast and safe. Confidence in shipping is high, even with evolving models and agent capabilities - Quality metrics (including model quality, agent reliability, and user experience) and dashboards provide clear, actionable signals to the team - You are a go-to partner for engineers, raising the bar for quality and reliability in AI-driven systems OUR TECH STACK We build on a simple modern stack optimized for both humans and AI. - Backend: TypeScript, Kotlin, Node.js - Frontend: TypeScript, React, Tailwind - Databases: Postgres, Redis - Cloud infra: GCP/Kubernetes/Terraform WHAT WE OFFER - Help shape AI's future alongside brilliant minds from Google, Apple, Spotify, Notion, Slack, Databricks, and BCG. - Competitive salary complemented with a competitive RSU program. - Swift professional growth in an evolving environment, supported by a culture of continuous feedback and mentorship from senior leaders. - Work with talented teammates across 5+ countries, and global customers, from our beautiful office in Stockholm, London and New York.