I
AI Engineer
InnerModal AI LLC
5 hours ago
Full-time
On-site
Salt Lake City, Utah, United States
Location: Remote (US); Salt Lake City / Salt Lake County preferred · Type: TBD · Team: Core team, Chief Architect
Build the place counselors practice before they meet real clients.
InnerModal AI builds several products on one core technology, Psychlones™. PatientProxy gives counseling trainees realistic AI clients and useful feedback for their supervisors. Trainees practice the hardest conversations safely, and their licensed human supervisor gets a short brief on where each trainee needs help. The AI never treats a client and never replaces a supervisor. We work smart, we work fast, and we have a lot of fun along the way. Responsibilities What you will do.
Own the platform end to end: LLM orchestration, persona rendering, session recording, scoring and the supervisor dashboard. Build evaluation into everything: test sets, human-agreement checks and regression tests after every model update. Make it safe for sensitive data: encryption, US hosting, access controls and audit logs. Interface with clients regularly, and recommend product changes based on their feedback. Help us choose models and keep costs in check. Outcomes
What success looks like.
First 30 days:
A market-ready Envoy for financial advisors: the bespoke digital spokesperson that speaks for advisors, never as them. By 3 months:
PatientProxy’s core loop (practice session → recording → scoring → supervisor brief) runs reliably end to end, and you’ve set up the evals that tell us whether a model change helps or hurts. By 6 months:
We’re demoing a market-ready PatientProxy to counseling programs, and you’re in the room with clients turning their feedback into the roadmap. By 12 months:
The platform handles sensitive data safely at real-customer scale, model costs are under control, and you’ve shaped how we build on Psychlones™ across products. There’s no single background that gets you here. We care more about what you can learn and build than about a checklist. If this sounds like the kind of puzzle you’d love, we’d love to hear from you. Bonus, not required.
Voice or real-time conversational AI. Experience with FERPA, HIPAA or other privacy-sensitive data. A background in psychology, counseling or education. Process
How we hire.
A paid work sample (about 6–8 hours): prototype one feedback feature and explain how you would test it. A conversation with our core team, and reference calls. Compensation commensurate with qualifications. We’re flexible about how and when you work: remote anywhere in the US, with hours that fit your life. We care about outcomes, not time at a desk. InnerModal AI is an equal opportunity employer. We welcome applicants of every background. Partner with us
Help put consistent, honest digital entities to work.
#J-18808-Ljbffr
InnerModal AI builds several products on one core technology, Psychlones™. PatientProxy gives counseling trainees realistic AI clients and useful feedback for their supervisors. Trainees practice the hardest conversations safely, and their licensed human supervisor gets a short brief on where each trainee needs help. The AI never treats a client and never replaces a supervisor. We work smart, we work fast, and we have a lot of fun along the way. Responsibilities What you will do.
Own the platform end to end: LLM orchestration, persona rendering, session recording, scoring and the supervisor dashboard. Build evaluation into everything: test sets, human-agreement checks and regression tests after every model update. Make it safe for sensitive data: encryption, US hosting, access controls and audit logs. Interface with clients regularly, and recommend product changes based on their feedback. Help us choose models and keep costs in check. Outcomes
What success looks like.
First 30 days:
A market-ready Envoy for financial advisors: the bespoke digital spokesperson that speaks for advisors, never as them. By 3 months:
PatientProxy’s core loop (practice session → recording → scoring → supervisor brief) runs reliably end to end, and you’ve set up the evals that tell us whether a model change helps or hurts. By 6 months:
We’re demoing a market-ready PatientProxy to counseling programs, and you’re in the room with clients turning their feedback into the roadmap. By 12 months:
The platform handles sensitive data safely at real-customer scale, model costs are under control, and you’ve shaped how we build on Psychlones™ across products. There’s no single background that gets you here. We care more about what you can learn and build than about a checklist. If this sounds like the kind of puzzle you’d love, we’d love to hear from you. Bonus, not required.
Voice or real-time conversational AI. Experience with FERPA, HIPAA or other privacy-sensitive data. A background in psychology, counseling or education. Process
How we hire.
A paid work sample (about 6–8 hours): prototype one feedback feature and explain how you would test it. A conversation with our core team, and reference calls. Compensation commensurate with qualifications. We’re flexible about how and when you work: remote anywhere in the US, with hours that fit your life. We care about outcomes, not time at a desk. InnerModal AI is an equal opportunity employer. We welcome applicants of every background. Partner with us
Help put consistent, honest digital entities to work.
#J-18808-Ljbffr