Sully.ai is hiring a Senior Software Engineer, Research

About Us At Sully.ai, We’re Building the Most Impactful Healthcare Company on Earth We believe that access to a great doctor is a basic human right. Today, that’s not a reality. Delays, misdiagnoses, administrative chaos, and burnout plague the system. Our Mission One Human, One Doctor. We enable our customers to staff 30% of their workforce with AI by creating a shared agent architecture for scale and efficiency. All powered by our own patented, world-class models and deployed in real-world care. What You’ll Do Build and optimize core research infrastructure: evaluation pipelines, agent workflows, hallucination detectors, coding benchmarks, and research→production integrations. Design, implement, and scale agentic systems across backend, frontend, and model integrations, collaborating closely with research and co-founders. Own reliability, observability, and performance across agents (logging, tracing, instrumentation, safety checks). Ship research-proven features into production within 7 days, end-to-end. Develop shared tools, SDKs, and internal products that accelerate iteration across Research, QA, and Engineering. Hard Requirements Senior-level full-stack engineering experience in React, TypeScript, and Node.js. Proven ability to design, ship, and scale LLM-powered applications. Expertise in API design, streaming, and CI/CD pipelines. Strong cloud infrastructure background (AWS, GCP, or Azure). Track record of building reliable systems with measurable performance and error budgets. First-Month Focus Audit all cross-agent flows for UI/UX consistency, correctness, and performance gaps. Implement shared components, typed schemas, and contract-driven interfaces for reliability. Establish instrumentation for frontend performance, agent consistency, latency, and model round-trip tracing. Improve or replace brittle evaluation or agent pipelines identified during onboarding. Partner with Research to productionize at least one new capability. 90 Day OKRs Deliver production-grade agentic workflows with