As a Principal Design Technologist within the Power Apps (App Agents) team, you will be a visionary leader at the intersection of design craft, AI architecture, and enterprise software. Your mission is to define and build the future of 'agentic experiences' for the Microsoft Power Platform, shifting from monolithic applications to composable, modular skills. You will be the primary design architect for how users and AI agents collaborate to build, automate, and innovate at scale.

Design Patterns & AI Semiotics (Primary Focus)

System Design: Deconstruct agentic AI systems into core functional subsystems (Reasoning, Perception, Action, Learning) and define the reusable patterns that govern them.

Interaction Semiotics: Architect the 'Sign Systems' for AI, defining how agents signal state, 'thought process,' and confidence. Establish standards for AI visual identifiers and conversational turn-taking.

Clarity of Thought: Establish requirements for 'Explainable Rationale,' ensuring every agent's action is grounded in a rule, preference, or prior user action.

Agentic Prototyping and Codebase Exploration

Vibe-Code to Production: Lead the rapid prototyping of 0-to-1 agentic features directly within the production codebase. Use AI-native development environments (e.g., GitHub Copilot, Claude Code, Lovable) to iterate on complex UI/UX patterns that require real-time model integration.

Functional Artifacts: Produce 'unsolicited provocations' and high-fidelity functional artifacts that challenge current product assumptions and reveal the technical constraints/opportunities of emerging LLMs.

Bridge Design and Engineering: Partner with professional developers to ensure that 'vibe-coded' prototypes are translated into scalable, maintainable, and high-performance production features through clear Pull Requests (PRs) and technical documentation.

Orchestration Design

MCP Architecture: Be familiar with the 'connective tissue' that allows agents to discover and utilize enterprise data across Dataverse, M365, and the Microsoft Graph.

Tool Design: Design and optimize MCP tool descriptions, instructions, and schemas. Ensure that agents can reason for 'when and how' to use specific business skills, maximizing intent fulfillment.

Pathway Design: Move beyond linear UI flows to design probabilistic 'pathways' for AI agents. Define how agents maintain memory, handle multi-step reasoning, and gracefully recover from non-deterministic failures.

Model Evaluation and Performance Frameworks

Quality Benchmarking: Define and operationalize the quality bar for agentic experiences by translating UX principles - helpfulness, coherence, and instruction fidelity - into measurable evaluation frameworks. Co-create 'LLM-as-a-judge' systems that embed these standards into continuous, automated evaluation loops, enabling scalable, high-confidence iteration of AI-driven experiences.

Ground Truth Creation: Collaborate with Research and Design partners to build 'ground truth' datasets and preference pairs for fine-tuning model behavior. Use qualitative user research to validate that agentic 'reasoning' aligns with human mental models.

Organizational Upskilling

AI Fluency Champion: Act as a force multiplier for the organization by upskilling designers, PMs, and cross-functional partners in AI tooling, prompt engineering, and agentic development patterns.

Design System Evolution: Drive the evolution of design systems into 'intelligent systems.' Ensure that all components, tokens, and icons are machine-readable and ready for agent-led composition.

Governance and Trust: Champion the principles of 'Responsible AI.' Ensure that all agentic experiences are built atop an enterprise-hardened foundation of governance, identity, and privacy.