Staff ML/AI Engineer

About the role

Lyra Health, a provider of evidence-based mental health care serving millions of people globally, is seeking a Staff ML/AI Engineer to define and drive the architectural vision for its machine learning and generative AI technology landscape. The role acts as a technical anchor across engineering and data, architecting enterprise-scale AI platforms, setting technical strategy for high-impact AI/ML initiatives and ensuring AI products deliver top-tier reliability, security and medical precision. It suits a seasoned technical leader who enjoys solving complex problems with cross-functional teams.

Key responsibilities

• Design and execute the long-term roadmap for the machine learning and generative AI platform, enabling fast, safe and reliable deployment of frontier models.
• Architect end-to-end training, fine-tuning and low-latency inference platforms, including centralized RAG architecture, vector databases and evaluation/guardrail frameworks.
• Establish standards for the full AI/ML SDLC, from dataset lineage and CI/CD pipelines to automated model evaluation, red-teaming and production monitoring.
• Partner with Product Management, Data Science, Security and Clinical leaders to turn clinical goals into AI capability roadmaps.
• Mentor Senior ML Engineers, run architectural reviews and set engineering best practices.
• Contribute hands-on through prototypes, critical-path architecture and strategic coding.

Required profile

• 8+ years deploying complex ML/AI solutions into mission-critical production environments, with a track record of technical leadership.
• Ability to turn ambiguous technical problems into clear strategic priorities and influence engineering, product and business leaders.
• Enjoys cross-functional work with physicians, therapists, data scientists, data analysts and product managers.

Required skills

• Python, RESTful APIs, Protobuf and microservices architecture.
• Docker, Kubernetes and real-time inference services.
• RAG pipelines, fine-tuned LLMs, vector databases and LLM guardrails.
• Relational databases, key-value stores, queues such as Celery and Kafka, and data pipeline orchestration.
• AWS or equivalent cloud architecture.
• Preferred: Java or Kotlin, HIPAA/SOC2 experience, MLOps and internal developer platforms.