Software Engineer, Agent Intelligence

About Sage Care

Sage Care is a fast-growing, early-stage healthcare startup founded by exceptional leaders from Apple, Uber, Carbon Health and backed by top-tier venture capital (General Catalyst, Chelsea Clinton). With a strong customer pipeline, Sage Care is transforming healthcare by simplifying care navigation.

Our platform makes it easier for patients to find the right doctor and helps providers focus on those who need them most through harnessing the latest AI innovations.

Building on our successful collaborations with health systems across the U.S., we have expanded internationally to the MENA region. We are now partnering with health systems there to deploy our AI-powered care navigation platform.

About the Role

Every day, our AI agents handle real patient conversations for hospital systems. Those conversations contain everything needed to make the agents better. Today, much of that learning is still manual: humans review calls, identify issues, investigate failures, and work with engineers to improve agent behavior.

Your mission is to build the systems that close that loop.

You will own the intelligence layer of our agent platform: the evaluation pipelines, feedback systems, and ML infrastructure that turn production conversations into measurable, continuous improvement.

This role sits at the intersection of AI evaluation, ML pipelines, and production quality. You will work closely with the engineers who own the agent runtime, with the operations teams who review calls, and with product leaders who decide what good looks like for each hospital partner.

What You'll Do

Build the evaluation and feedback platform

Design systems that continuously analyze production conversations, identify and cluster quality issues

Build evaluation pipelines that measure agent performance across the dimensions that matter for patient care

Develop workflows that transform human feedback into actionable improvements

Design mechanisms for routing issues to the appropriate AI, engineering, or operational owners

Make failures diagnosable

Investigate production failures and identify root causes across transcription, reasoning, retrieval, and orchestration

Build tooling that helps engineers quickly understand why an agent behaved a certain way

Establish quality metrics and reliability standards for production agents

Automate the learning loop

Build ML pipelines that reduce the manual effort required to improve agents

What We're Looking For

Required

5+ years of software engineering experience

Experience building production systems

Experience working with LLMs, AI agents, or conversational AI applications

Experience in one or more of the following:

building AI evaluation platforms or frameworks

developing feedback systems for ML and LLM applications

creating observability, reliability, or quality tooling for AI products

building ML pipelines that improve model or agent performance

Strong backend engineering skills and systems thinking

Experience working with ambiguous problems and defining solutions from first principles

Nice to Have

Experience with evaluation frameworks and model quality measurement at AI-native companies

Experience with voice AI systems

Experience designing human-in-the-loop workflows

Experience turning emerging research or new techniques into tested prototypes

Experience building internal platforms used by engineering teams

Experience in healthcare or other regulated, safety-critical domains

What Success Looks Like

Failures are detected by systems, not discovered by people

Every production issue has a diagnosable root cause and a clear owner

Human feedback measurably changes agent behavior, with the lag between the two shrinking over time