Remote Ai Agent Developer Jobs in Vermont (NOW HIRING)

Reports to: Head of Product

Location: Burlington, VT (hybrid) or remote

OhMD is an AI patient communication platform used by thousands of healthcare organizations. Our product suite includes two-way patient texting, an AI voice and text assistant that handles inbound patient calls with a human always in the loop, and integrations with 85+ EHRs. Practices using OhMD reduce inbound call volume by 50% or more which is the difference between a front desk that is drowning and one that can actually take care of patients.

OhMD is patient communication software for physician practices. Two-way texting, digital forms, automated workflows — and OhMD AI, our AI voice agent. It picks up on the first ring, handles the routine calls your front desk shouldn't have to take, and hands the conversation to a human the second it needs one, with full context attached. Practices running OhMD see 68% fewer calls reach their staff.

That handoff is the whole point. We've never believed you can automate a truly human interaction. The best AI available today can confidently resolve maybe 60% of what a practice hears in a day. The other 40% is where the nuance lives — an anxious parent, a complicated refill history, a patient who needs to be heard rather than routed. Our whole product philosophy is knowing the difference.

Which brings us to the hard part, and to you.

Every practice is different. A pediatric group handles refill requests nothing like a dermatology practice. Their EHR is configured differently. Their front office has its own unwritten rules about what gets escalated, to whom, and how fast. Our AI has to learn all of it — and be right, live, on a real call, with a real patient.

This role owns that. You'll define how OhMD AI behaves for the practices we bring live, build that behavior yourself, and stay on it until the calls resolve cleanly.

• Weeks 1–4: Live in the product. Listen to calls. Shadow go-lives. Learn how a practice actually runs and where the AI currently frustrates people.
• Weeks 5–8: Own your first configurations end to end, with support. Ship changes to live agent behavior.
• Weeks 9–12: Own go-lives independently. Bring us a point of view on what's breaking at scale and what we should build next.

How the AI behaves in the real world. You'll sit with the people who actually answer the phones, learn their workflows, and turn what you hear into working agent behavior — conversation flows, decision logic, escalation rules, edge case handling. Not a spec you hand to someone else. You'll build it, in our platform, with prompts, tools, and guardrails.

Practice go-lives, start to finish. Our practices go live in about three weeks: platform setup, EHR integration, AI configuration, staff training. You're the single owner of what the agent does across that window — scoping what's in and out, making the calls when something doesn't fit the standard path, and making sure the practice's first week live is a good one.

Quality, measured honestly. You'll listen to real calls. A lot of them. You'll build evals that catch failures before a practice does, read the transcripts where the agent got it wrong, and close the loop. When you find a failure mode our platform can't handle yet, you'll write the requirement and work it through with engineering.

Depth in healthcare's messiest systems. We integrate with 85+ EHRs — athenahealth, eClinicalWorks, Epic, AdvancedMD, ModMed, and a long tail after that. Scheduling logic, patient context, and refill data all come from somewhere, and that somewhere is rarely tidy. You'll scope those integration requirements alongside our engineering team and design agent behavior that holds up when the data doesn't cooperate.

Patterns that outlive you. The first time we solve a workflow, it's a project. By the fifth time, it should be a template. You'll notice when something's ready to graduate from custom work to repeatable setup, write it down, hand it to the implementation team, and move on to the next unsolved thing.

• Representing OhMD externally. Occasional appearances at industry conferences, webinars, or partner events (Twilio, EHR partners) to talk about how the AI agent actually works in practice.
• Helping build the team. Sitting in on interviews for future AI PM or implementation hires as the team grows, since you'll know better than anyone what "good" looks like for this seat.
• Other duties as assigned.

• 3+ years in product management or an adjacent seat — technical program management at a product company, a founding-team role where you wore the PM hat, or an engineering background where you've been the one talking to customers and deciding what gets built.
• Technical enough to be dangerous. You're comfortable with APIs and integrations and you can reason about how systems connect. You don't need to ship production code, but you shouldn't need a translator.
• Hands-on instincts. When you want to know if something works, you go build a version of it.
•  Comfort with ambiguity. Much of this work has no playbook yet. You'd rather make a decision and correct course than wait for someone to hand you clarity.
•  Influence without authority. You'll coordinate across engineering, implementation, and customer success without managing any of them.
• Care about the end user. The patient on the other end of that call is often stressed, sick, or worried about someone they love. That should matter to you.
• Hands-on work with LLMs, AI agents, or prompt engineering. Extra credit if you've broken a few on purpose to see what happens

• Background in healthcare, or another industry where the underlying systems are old and unforgiving
• Experience with voice AI or real-time conversational systems
• Examples of conversational experiences you've shipped — voice, chat, SMS, IVR — and the reasoning behind the choices you made
• Familiarity with evaluating agent quality in a structured way
• Early-stage startup experience