Refactor Verified
Founding AI Engineer
Palo Alto, California, United StatesOnsiteFull Time$120,000–$150,000 /yrPosted 27 days agoHidden Gem · YC Startup
Role summary
Refactor is seeking a Founding AI Engineer to join their team in Palo Alto. This role involves building a configurable AI platform to automate healthcare reimbursement processes. The engineer will focus on creating shared infrastructure and tooling, developing autonomous loops for failure detection and resolution using LLMs, and working with diverse data modalities. The position requires strong backend (Python, FastAPI, PostgreSQL) and frontend (TypeScript, React, Next.js) skills, along with experience in AI/ML technologies like fine-tuning, distillation, and LLM unit-testing. The role also involves direct engagement with healthcare companies.
Refactor is working to eliminate the $50B+ labor cost of healthcare reimbursement.\
\
We’re building a configurable AI platform that works with existing electronic health record systems to enable autonomous billing for healthcare providers.
We are extremely fortunate to be working with enterprise billing companies handling $4B+ in medical claims across hundreds of clinics along with some of the largest health systems in the country ($10B+ in net patient revenue), which has given us a great position to attack and solve this critical problem for the largest and most important vertical in the world.
### **Interesting Technical Problems**
Healthcare reimbursement is incredibly complex due to the large number of insurance companies (and third-party administrators), electronic health record systems, and specialties, each with their own rules and workflows. This makes deployment and configuration to each of our tenants exceptionally cumbersome and time-consuming.
An immediate goal of ours is to create shared infrastructure and tooling to automate this (think of this like an AI FDE), which has proven to be a difficult and exciting technical challenge!
As a starting point, we’ve created an autonomous loop that detects failures (real and in simulation), creates llm unit-tests to reproduce those failures, and iterates until those are solved without introducing regressions. This is rather effective at autonomously tuning the behavior of our agents without significant manual interference. We hope to improve and push the boundaries here significantly.
Another exciting fact is that we are building systems that have to work across diverse modalities: pdfs/faxes, phone calls, web portals, desktop apps, and more, which means that you will get exposure to an extensive breadth of techniques and challenges.
### **Job Description**
We will be working in person in Palo Alto. You will be spending 70-80% of your time writing maintainable and performant code (heavily enabled by all the relevant AI tooling) and the remaining time working hands-on with critical healthcare companies across the country.
Backend: Python FastAPI, PostgreSQL
Frontend: TS (React + Next.js)
Technologies: Temporal, custom eval infra (inc. fine-tuning and distillation), LiveKit, etc.
\
We’re building a configurable AI platform that works with existing electronic health record systems to enable autonomous billing for healthcare providers.
We are extremely fortunate to be working with enterprise billing companies handling $4B+ in medical claims across hundreds of clinics along with some of the largest health systems in the country ($10B+ in net patient revenue), which has given us a great position to attack and solve this critical problem for the largest and most important vertical in the world.
### **Interesting Technical Problems**
Healthcare reimbursement is incredibly complex due to the large number of insurance companies (and third-party administrators), electronic health record systems, and specialties, each with their own rules and workflows. This makes deployment and configuration to each of our tenants exceptionally cumbersome and time-consuming.
An immediate goal of ours is to create shared infrastructure and tooling to automate this (think of this like an AI FDE), which has proven to be a difficult and exciting technical challenge!
As a starting point, we’ve created an autonomous loop that detects failures (real and in simulation), creates llm unit-tests to reproduce those failures, and iterates until those are solved without introducing regressions. This is rather effective at autonomously tuning the behavior of our agents without significant manual interference. We hope to improve and push the boundaries here significantly.
Another exciting fact is that we are building systems that have to work across diverse modalities: pdfs/faxes, phone calls, web portals, desktop apps, and more, which means that you will get exposure to an extensive breadth of techniques and challenges.
### **Job Description**
We will be working in person in Palo Alto. You will be spending 70-80% of your time writing maintainable and performant code (heavily enabled by all the relevant AI tooling) and the remaining time working hands-on with critical healthcare companies across the country.
Backend: Python FastAPI, PostgreSQL
Frontend: TS (React + Next.js)
Technologies: Temporal, custom eval infra (inc. fine-tuning and distillation), LiveKit, etc.
Similar roles
- Founding AI EngineerMage Legal · San Francisco, California, United States · Onsite
- Founding AI EngineerPingo AI · San Francisco, California, United States · Onsite
- Founding AI EngineerEmbedder · San Francisco, California, United States · Onsite
- Founding AI EngineerKinro · San Francisco, California, United States · Onsite
Founding AI EngineerRaya Health · San Francisco, California, United States · Onsite
