Senior Software Engineer
@ Flagler Health
April 2025 — Current
- Built and tuned AI systems across traditional predictive models and LLM-based agents, using LLMs as part of the model-tuning loop while maintaining distinct training, evaluation, context, and state-management workflows.
- Developed and deployed 30+ composable AI skills for engineering and operations, combining them into higher-level workflows with structured context and state management so agents can reason and act against real application state, not isolated prompts.
- Made AI-assisted implementation verifiable by encoding human-reviewed reasoning into repeatable checks, then applying adversarial review and production-data validation to catch failures conventional tests missed.
- Helped shape the voice-agent layer, connecting patient context, scheduling, clinical workflows, and operational actions to bring conversational AI directly into care delivery.
- Hydrated 90M+ healthcare records and data points at query time across claims, payers, providers, clinics, and EHR/EMR systems, reconciling inconsistent source data into structured inputs that users, automation, and AI can reason over.
- Turned fragmented clinic operations into structured, queryable primitives, establishing an extensible data model and workflow rails that support new use cases without requiring another bespoke integration.
- Developed an agent-compatible task and action execution layer spanning 70+ task types and 400+ configured triggers, delivering 100+ trigger executions/sec/container through event-driven processing, composable workflow logic, dynamic entity references, and end-to-end task lifecycle automation.
- Codified clinic tribal knowledge and ambiguous SOPs into deterministic, auditable medical-necessity and coverage rules across 75+ clinics and 2,000+ plan configurations, spanning 22K+ payors and 19K+ known-good rules.
- Delivered a time-tracking pipeline currently processing 21M+ care delivery actions, enabling clean reconciliation of enrollment, patient engagement, and EHR activity into reliable billing and operational analytics.
- Established engineering guardrails for AI-native development, helping engineers adopt AI without shifting the entire bottleneck to code review and validation.
- Shaped foundational platform capabilities around determinism: making state explicit, reasoning inspectable, and behavior reproducible by both humans and AI.