> Markdown version of [/jobs/ext/3474657-founding-software-engineer](https://www.wearedevelopers.com/jobs/ext/3474657-founding-software-engineer). Every page supports `.md` or `Accept: text/markdown`. Links point to the HTML versions so they work for humans too. Agent guide: [/agents.md](https://www.wearedevelopers.com/agents.md). --- # Founding Software Engineer - **Company:** KERNEL - **Location:** New York, NY, United States - **Salary:** $100,000.0 - $120,000.0 - **Contract:** Permanent contract - **Skills:** Training Data, Artificial Intelligence, Software Debugging, Medical Software, Alwayson, Reinforcement Learning, Stack Overflow - **Published:** September 21, 2026 - **Apply:** https://www.indeed.com/viewjob?jk=0d2393dd1a3ded94 ## About the Role We care much more about what you can build than where you went to school or how many years you have worked. You may be a strong fit if you: * You are an exceptional programmer * You have unusually high technical ambition * You learn new systems very quickly * You have built serious projects outside of assigned work * You enjoy debugging problems where there is no Stack Overflow answer * You are comfortable moving between ML, infrastructure, product, and systems work * You want ownership rather than narrowly scoped tickets * You care about both speed and engineering quality * You want to work intensely with a small team in person in New York * You enjoy making things We are open to exceptional new graduates and early-career engineers. ## Description Healthcare providers spend enormous amounts of time navigating insurance portals, practice-management systems, phone calls, claims, denials, payments, and reconciliation just to get paid for care they have already delivered. We think much of this work can and should be done by machines. Kernel is an always-on system that operates across the software healthcare practices already use. Our agents navigate practice-management systems and payer portals, call insurance companies, verify coverage, submit and recover claims, reconcile payments, and increasingly handle the entire path from care delivered to money collected. We are starting with dental practices and building toward a much larger goal: making healthcare payments simple. The role We are looking for an exceptional engineer to join the founding team in New York. You will work directly with the founders and own major pieces of Kernel from idea to production. This is not a narrowly scoped engineering role. You might spend one week building an agent that operates legacy healthcare software, the next improving our verification infrastructure across thousands of sessions, and the next shipping a new payments product directly to clinics. We are looking for an exceptional engineer to join the founding team and help us push these systems toward true autonomy. What you'll build You will work directly with the founders on some of the hardest technical problems at Kernel. This may include: * Building computer-use and browser agents that reliably operate legacy healthcare software * Training and improving models for healthcare-specific actions and reasoning * Building post-training pipelines using SFT, reinforcement learning, synthetic data, and production traces * Designing rigorous evals for long-horizon agent behavior * Improving reliability across workflows that may require hundreds of actions and multiple external systems * Building model-routing, inference, context, memory, and tool-use infrastructure * Turning millions of real-world agent interactions into training data * Building voice agents that communicate with insurance companies * Working with multimodal models across screens, documents, faxes, forms, and structured data * Reducing inference cost and latency while increasing task success * Building the product and infrastructure necessary to deploy these systems across thousands of healthcare practices, Milo is a small engineering team in Manhattan. Engineers here do not sit several layers away from the problem. You might talk to a customer in the morning, inspect an agent failure, modify the runtime, run an experiment, and deploy the fix that afternoon. We expect engineers to own problems end to end. There are very few meetings, very little bureaucracy, and a very high bar for what we build. Why Milo AI is becoming capable of doing work, not just generating text. Healthcare is one of the largest industries in the world, but much of its financial infrastructure still depends on people manually moving information between fragmented systems. We think that changes over the next decade. Our goal is to build the autonomous infrastructure that gets healthcare providers paid. If that sounds like the kind of problem you want to spend the next few years solving, we'd like to meet you.