Led design of control planes for agentic workflows enabling developers to deploy, configure, evaluate, and safely expose LLM and foundation model behavior to production traffic through shadow testing and staged rollout.

Built functional design to code prototypes daily using Kiro and Claude-based AI tools, exploring and validating complex workflow designs directly in code alongside engineering partners.

Conducted user interviews and usability testing with engineering partners to validate control-plane workflow designs before production rollout.

Collaborated with Science and Engineering teams to define interaction patterns for agentic experiences, reviewing technical documentation and agent behavior prototypes to translate system capabilities into usable end-user flows.

Led UX design for a real-time observability platform used by engineering and operations teams to diagnose and resolve issues in high-volume streaming environments.

Transformed high-cardinality telemetry data into structured, navigable dashboards that supported rapid root cause analysis and decision-making under time pressure.

Grounded workflow design in field research and support channel analysis to accurately model SRE diagnostic patterns.

Led design for AI-assisted career exploration features within a large-scale enterprise platform.

Defined interaction patterns that introduced machine learning and generative AI responsibly into user workflows, emphasizing explainability, user control, and enterprise-grade reliability.

Worked cross-functionally with product, engineering, and data science to align AI capabilities with compliance, security, and system performance requirements.

Mentored designers and contributed to cross-team design standards.

Led end-to-end UX for AWS Glue Studio, a distributed ETL infrastructure tool for data engineers and scientists, driving 7x user adoption within six months through research-informed workflow design.

Designed DAG-based pipeline authoring, embedded SQL transform interfaces, and runtime monitoring views with full operational control including modify, stop, and rollback workflows with guardrails to prevent invalid system states.

Led end-to-end design of a query editor for Amazon QLDB with PartiQL support, type-ahead, keyboard shortcuts, multi-statement execution, and onboarding flows for technical users adopting NoSQL data structures.

Led design for a virtual assistant platform used across internal and partner systems.

Established reusable conversational patterns and escalation models that balanced automation with transparency, ensuring users understood system boundaries and next steps.

Contributed to shared design components supporting scale and cross-team consistency.