Senior AI Engineer

Florence software advances cures by helping the world’s most important research sites do their best work. Our solutions are now used by over 30,000 research teams in 70 countries around the world—we’re the most widely deployed site workflow tool in the industry. By the end of the decade, we’ll double the pace at which new medicines get to market by doubling the output of trial site teams.

To date, we were named a Deloitte Fast 50 business, G2 Category Leader, an Inc. & AJC best place to work, and an Inc. 5000 company five years in a row.

At Florence, we are committed to make the world a better place by accelerating research while providing an environment for our employees where they can be happy in their lives, enjoy their jobs, and grow.

We are looking for a Senior AI Engineer to design, build, and deploy high-quality AI-powered features. This role focuses on owning end‑to‑end implementation of AI systems within a product area — from prototyping to production — with a strong emphasis on reliability, iteration, and measurable impact. You will work closely with product and engineering teams to turn ambiguous problems into effective AI solutions, while contributing to best practices and raising the bar for AI development.

• Design and implement AI-powered features (LLM workflows, copilots, and agent-based systems with tool use and multi-step reasoning)
• Ensure solutions are reliable, performant, and aligned with product needs

• Build and optimize prompt pipelines for specific use cases
• Retrieval systems (embeddings, chunking, ranking)
• RAG-based workflows where needed
• Iterate on outputs to improve quality, accuracy, and consistency
• Design scalable and cost‑efficient AI architectures for production workloads
• Select and evaluate models (hosted vs open‑source) based on use case constraints
• Agent-Based Systems (Agent Core)
• Design and build agentic workflows capable of multi-step reasoning and decision-making
• Integrate agents with tools, APIs, and internal systems to perform real-world actions
• Implement planning, execution, and reflection loops for complex tasks
• Manage context, memory, and state across multi-step interactions
• Balance deterministic workflows vs. agent autonomy for reliability and control
• Run structured experiments to compare approaches (prompting, retrieval, models)
• Define and track key metrics for AI performance (quality, latency, cost)
• Debug and improve non‑deterministic system behavior
• Build and maintain evaluation datasets and benchmarks
• Implement automated evaluation pipelines for continuous improvement
• Drive technical direction and influence AI adoption across teams
• Partner with product managers and designers to scope AI features
• Contribute to shared patterns and reusable components
• Participate in code reviews and design discussions
• Support and mentor mid‑level engineers where needed

• Design guardrails to ensure safe and reliable AI behavior
• Mitigate hallucinations, prompt injection, and model misuse
• Ensure compliance with data privacy and enterprise requirements
• Implement monitoring and observability for AI systems in production
• Implement guardrails for agent actions (tool access control, execution boundaries)
• Prevent failure cascades in multi-step agent workflows

• Strong understanding of LLM capabilities and limitations
• Experience with prompt engineering and structured output design
• Hands‑on experience with embeddings and vector search
• Familiarity with RAG architectures and when to apply them
• Experience designing agent-based architectures (Agent Core concepts)
• Understanding of tool use, planning strategies, and memory mechanisms in LLM systems
• 4+ years of related work experience
• Evaluate agent performance across multi-step tasks (task success rate, error propagation)
• Debug and optimize agent decision-making and tool selection behavior
• Experience building and deploying production‑grade systems
• Ability to debug complex issues, including probabilistic outputs
• Comfort working with APIs, pipelines, and data flows

• Ability to translate user needs into…