Senior Software Engineer, Platform

Stelloagents; Northern (KY); Hybrid; USD 140,000 - 190,000; Full time

Job summary

Stello, an AI transformation firm, seeks a Senior Software Engineer for the Platform team to own orchestration, evaluations, observability, and safety guardrails for production agents. You’ll shape architecture, drive reliability, and raise the bar for engineering excellence, partnering with Forward Deployed Engineers to deploy scalable foundations for client work.

You will design long-running orchestration, build evaluation systems, and provide run traces, metrics, and alerts.

Qualifications

• 5+ years building production software that people depended on, and supporting it after launch.
• Sound judgment about architecture, testing, reliability, and maintainability.
• Experience with distributed systems, queues, or workflow engines.
• You have built with LLMs, and you can explain how you measured and improved their output.
• You use coding agents every day, and you know how to review and validate what they write.

Responsibilities

• Design and build orchestration for long-running, multi-step agents.
• Build the evaluation system that tells us if a change to a prompt, a model, or a tool made an agent better or worse.
• Give every agent run traces, metrics, and alerts, so we see a problem before a client does.
• Build controls: permissions, rate limits, human review steps, and safe ways to fail.
• Turn patterns from client work into components the delivery team can reuse.
• Make recommendations, explain your reasoning, and raise issues early when a plan needs to change.

Job description

About Stello

Stello is an AI transformation firm. We build AI agents that take over operational workflows for mid-market and enterprise companies. Our clients range from professional sports teams to public insurers and financial services firms. Some of them trade on the NASDAQ and NYSE.

The role

As a Senior Software Engineer on the Platform team, you own the system that every Stello agent runs on: orchestration, evaluations, observability, and the guardrails that keep agents safe in production.

Our Forward Deployed Engineers build agents for individual clients. You build the shared foundation under them, so each new deployment ships faster and breaks less. You own the platform from architecture to production performance, and you set the engineering bar for the team.

What you will do

• Design and build orchestration for long-running, multi-step agents.
• Build the evaluation system that tells us if a change to a prompt, a model, or a tool made an agent better or worse.
• Give every agent run traces, metrics, and alerts, so we see a problem before a client does.
• Build controls: permissions, rate limits, human review steps, and safe ways to fail.
• Turn patterns from client work into components the delivery team can reuse.
• Make recommendations, explain your reasoning, and raise issues early when a plan needs to change.

Problems you will work on

Agents are hard to make reliable. Data is fragmented, client systems are old, and model behavior is difficult to predict. Many of the engineering patterns for agents are still being worked out.

• Retry a ten-step agent run that failed at step seven, without repeating the side effects of the first six steps.
• Prove that a new model version is safe to roll out to every client.
• Give agents useful memory across long tasks, without leaking data between clients.
• Show a client's operations lead what an agent did today, in words they understand.

Improve how we build

We use coding agents across planning, implementation, testing, and review. You will shape how we break work into tasks for agents, how we give them context, how we check what they produce, and where a person or plain code is the better choice.

What we look for

• 5+ years building production software that people depended on, and supporting it after launch.
• Sound judgment about architecture, testing, reliability, and maintainability.
• Experience with distributed systems, queues, or workflow engines.
• You have built with LLMs, and you can explain how you measured and improved their output.
• You use coding agents every day, and you know how to review and validate what they write.
• You take responsibility for delivery, and you say early when something needs to change.

We care about the quality of your work, your judgment, and how fast you learn. Years of experience are only one part of that.

How we work

This is a full-time remote role, open to candidates who are authorized to work in the United States. You report directly to the founders. You may travel to client sites when a project calls for it.

We are a small team. We review each other's work, share what we learn, and help when a teammate is stuck. People who deliver get more scope quickly.