Senior Software Engineer

Every Keystroke You Spend Writing Code Is One That Could Be Better Spent Writing Instructions For Your Army Of Agents

A few weeks ago a 95-year-old patient called one of our practices. She apologized for being long-winded before she'd said much of anything. "I promise my mind is still very sharp," she said; you could hear in her voice that she'd been made to feel otherwise. Our agent didn't cut her off. It listened. She thanked it at the end of the call for being so patient.

She probably didn't know she was talking to AI. But she felt that someone had listened.

Most voice agent companies in healthcare are racing to solve one thing: getting patients off hold fast enough to schedule an appointment. We think that's the wrong finish line. The moment a patient calls is just the beginning: there's the referral fax, the insurance check, the prescription refill, the follow-up, the billing. Most of it is invisible to the patient and exhausting for the practice. We're building the platform that handles all of it, with the patient at the center of every touchpoint.

We're 2 years in and working with 400+ medical practices across 27 specialities, including primary care, dermatology, optometry, and mental health. One Medical's founder is an investor. We've raised over $13M in VC funding and grew ARR 10x last year, which means it's a great time to join.

The Role

Honest take: the tools we're building with today could be obsolete in six months, maybe sooner. The engineers who thrive here are the ones who find that energizing, and who treat LLMs as force multipliers for everything: thinking through problems, catching edge cases, moving faster than should be possible.

In your first two weeks, you'll launch a new end-to-end agent use case for a live customer. Not a toy project; a real one, for a practice that serves millions of patients. The gray-area decisions around patient experience are the ones LLMs consistently fumble: what the agent says when something goes wrong, how it recovers mid-conversation, when it hands off to a human. Someone with actual judgment has to own those calls, and that someone is you.

Your job, more broadly, is to harness LLMs to give yourself as much leverage as possible across everything you do. Use them to uncover product gaps, pressure-test your own ideas, and compress the distance between a thought and a shipped thing. (Use them to order your lunch if that's what it takes. We mean that. Any time spent increasing leverage while pushing the quality bar higher is time well spent here.)

You'll also be closer to customers than most engineers at your level, because the best product decisions here come from engineers who heard the problem firsthand. No feedback filtered through three layers of telephone.

What it looks like day to day: our engineers plan and build using an internal agent harness with custom skills. We already have agents crawling our Datadog logs and fixing bugs autonomously. Non-engineers are one-shotting bug fixes in Slack (I'll be honest, I was skeptical of this until I watched it happen). Even our onboarding docs are agentic: we have a custom agent that conjures up a design diagram in your favorite color on demand. Getting to this point required a lot of upfront investment, including building out our harness with rich context graphs and MCPs, and we're starting to enjoy the fruits of that. But we've only scratched the surface, and we know it.

Our North Star is that an engineer here could ship meaningful work without manually writing a single line of code… maybe ever. We're probably 15% of the way there today, and we think we could be at 70% in four months. Getting us there is part of the job.

What makes this harder than it sounds: our agents aren't handling toy interactions. They're speaking with real patients—people who are afraid, in pain, or navigating something complicated—and what we ship runs on their most sensitive health data. A fast-moving, LLM-heavy engineering culture has to coexist with a genuinely low margin for error. We've spent a lot of time figuring out how to make that work. The short version: it's possible, it's working, and there's still a lot of runway ahead.

What You'd Work On

One of our customers was processing 50,000 faxes a month. Each one required a person to spend five minutes reading, sorting, and inputting the relevant information—armies of 10 to 20 people, eight hours a day, five days a week, just to keep up. The day after they brought this problem to us, we had them forwarding faxes into our system. Within two weeks we had an MVP running on actual production faxes: a tool that reads incoming referrals and prior authorizations and calls the patient to schedule them. What used to take 5 to 7 days now takes one minute. That product is live today, and we didn't cut corners on compliance or quality to get there.

Building on and expanding that platform is the work. Specific problems you'd wrestle with:

Voice and conversation design at scale. Our agents speak to tens of thousands of patients daily. The gap between "technically correct" and "actually human" is enormous, and it's not a problem you can prompt your way out of. One of our engineers spent three or four hours generating iterations of filler phrases ("mmm, let me look that up for you") because cadence and intonation matter to a patient calling about their health. An LLM can generate the options. Someone with ears has to choose.

Knowing when the agent should stop. Healthcare conversations move through predictable stages—intake, verification, scheduling, follow-up—and what's appropriate to say at each stage isn't always obvious from the outside. A patient's diagnosis might be relevant at one point in the conversation and completely out of bounds thirty seconds earlier. A model optimizing for correctness will flatten all of that. Building systems that know not just what to say but when—and that route edge cases to a human without grinding everything to a halt—is some of the most consequential work we're doing. Get it wrong and it's not just a bad experience. Depending on what gets said to whom, it can be a compliance violation.

Building the infrastructure that makes all of this possible. Getting an agent to fix a bug autonomously sounds simple. Getting it to do that reliably, on a production codebase that handles real patient data, is a different problem. It requires evaluation systems that can tell the difference between a good fix and a confident wrong answer, context that's rich enough for the agent to understand what it's touching, and guardrails that fail safely when it doesn't. We've built a lot of this. There's more to build. And the honest truth is that the agents are only as trustworthy as the humans who designed the systems around them.

A Day in Your Life

No two days look the same, but here's a realistic sketch:

Morning: You're building out a new capability on the agent platform: a way to handle prescription refill requests that a practice flagged as fully manual and error-prone. You're using the internal agent harness to scaffold the initial implementation, then reviewing what it produced and making the judgment calls it couldn't. If you need to test something under real-world conditions—noise reduction, how the agent handles a difficult accent—you might head to the cafe downstairs. (We get out of the lab.)

Late morning: An alert surfaces from the Datadog agent: something's off in how insurance information is being collected in a specific scenario. The symptom points to one place; the root cause turns out to be somewhere completely different. The agent had context across the entire codebase, meeting transcripts, and Notion pages and traced it faster than any engineer who knew the stack. You verify the fix and ship it.

After lunch: One of the QA folks who audits patient calls posts in Slack that she needs a "Mark as Reviewed" button. It takes a few seconds per call, but she reviews thousands of them. She tags the Slack bot hooked to our codebase. Five minutes later the button exists. You give it a once-over and it's live.

End of day: You sync with CSMs and Eric on what they heard from customers. One conversation surfaces a new product opportunity. You sketch what a prototype might look like. (You'll be building it tomorrow.)

Who Thrives Here

You treat LLMs as force multipliers and you're already working that way.

You have taste. It's the thing that's hardest to hire for and most