Phare is building healthcare’s first Revenue Operating System that uses AI to make hospital billing and reimbursement effortless, accurate, and fair. The company is looking to hire an MLOps Engineer to own the production runtime for their ML stack.
Requirements
- Production ML: Deployed and operated models running on GPUs in production - APIs and batch/streaming inference.
- Platform engineering: Strong with Docker/Kubernetes, IaC (e.g., Terraform), and CI/CD for services and model artifacts; maintain environment parity, reproducible releases, and robust model/experiment versioning with data lineage.
- System Reliability: Use progressive delivery with automated rollouts/rollbacks, and build end-to-end observability (metrics, logs, traces, and model telemetry for drift/regression) plus actionable alerting, runbooks, and incident response.
- Post-training lifecycles: Manage model registries and stage gates, design scheduled or event-driven retraining when appropriate, and enforce RBAC, secrets management, encryption, and audit logs.
- Experience in regulated environments (e.g., healthcare, finance)
- 5 years of software engineering experience
- 2 years of MLOps experience
Responsibilities
- Deploying, serving, and scaling models across inference endpoints and batch/streaming workflows.
- Build progressive delivery pipelines with automated rollouts and rollbacks.
- Manage SLOs for latency and availability.
- Instrument end-to-end observability (metrics, logs, traces, drift, regression).
- Harden the platform with Terraform, Kubernetes, and CI/CD, ensuring reproducible, auditable ML releases.
- Operate ML systems at scale, where uptime and feedback loops matter as much as accuracy.
- Deploy and operate models running on GPUs in production - APIs and batch/streaming inference.
Other
- This is an in-person role in NYC requiring at least 3 days in the SoHo office.
- Hybrid in-office (min. 3 days per week)
- Intro call - your background & our mission alignment
- Culture interview in person in NYC
