Senior Staff Software Engineer (AI Team)

Splose is an AI-powered practice management platform focused on improving healthcare for clinicians. They are seeking a Senior Staff Software Engineer to build and evolve AI features that enhance clinician workflows, including voice transcription and intelligent note generation, while ensuring reliability and security in AI integrations.

Responsibilities:

• Build and maintain AI features end to end, including voice and audio pipelines, LLM integrations, embedding and vector search infrastructure, and the clinical facing AI surfaces built on top of them
• Solve AI specific reliability and quality problems such as context window management, rate limiting, streaming failures and model upgrades, with solutions that stay transparent and trustworthy to the practitioner
• Treat security as part of standard AI delivery: scope every AI query to the organisation, prevent prompt injection, handle audio and file data securely, and stay ahead of model and dependency vulnerabilities
• Design and ship in-product systems that drive AI adoption, including onboarding experiences, usage nudges and experiments that help clinicians form habits with AI features
• Establish comprehensive observability and data lineage tracking for AI workflows, and define and monitor SLIs/SLOs (token usage, latency, error rates) to catch degradation before it reaches clinicians
• Act as the technical standard bearer for the squad by enforcing rigorous code reviews, defining paved paths for architecture, and establishing engineering practices that prioritise reliability and maintainability
• Drive continuous improvement in squad delivery metrics, particularly Lead Time for Changes, by shipping voice pipelines and LLM integrations with deterministic automated tests that validate system boundaries before deployment
• Meet or exceed defined SLOs and error budgets for all AI components, and reduce Mean Time to Recovery for AI specific incidents through robust observability and actionable runbooks
• Define and enforce clear architectural paved paths that reduce the cognitive load required for other engineers to safely interact with non-deterministic AI components
• Maintain a zero-incident record on cross-tenant data bleed by mitigating prompt injection risks and data isolation vulnerabilities at the architectural design phase, not after release
• Partner with the Principal Security Engineer on threat models and controls, and with the Head of Engineering Enablement on resilience models and controls