Artificial Intelligence

I don’t just follow AI trends or spout buzzwords – I lead practical, scalable AI integrations that solve real workflow problems. And I have the education and experience to offer a deep understanding of all steps of gathering and cleaning data, creating the model, assessing the confidence and potential weaknesses of the model, and making the user interface. And I’m passionate about making it ethical and human-centric, too.

Most recently, I led the exploration and adoption of Figma’s new AI features within Edward Jones’ design organization. As the first designer to champion the opportunity, I owned the process from initial proposal to pilot rollout, ensuring we weren’t adopting AI for AI’s sake, but rather to meaningfully reduce design debt, speed up documentation, and support non-visual contributors.

Edward Jones Chatbot

I provided design and best practices feedback for Edward Jones’ new chatbot, including:

Conversational Design Best Practices

  • Reviewed tone, structure, and phrasing for clarity, empathy, and professionalism

  • Ensured concise responses, appropriate fallback messages, and error handling UX

  • Recommended progressive disclosure (don’t overwhelm the user all at once)

  • Identified opportunities for more natural language triggers or entry points

Information Architecture & Flow Logic

  • Audited conversation flows for redundancy, circular loops, or dead ends

  • Recommended clear entry/exit paths and actionable handoff moments

  • Suggested decision-tree refinements to match common user intents

Accessibility & Inclusive UX

  • Reviewed color contrast, keyboard navigation, screen reader support

  • Flagged potential cognitive load issues, unclear labels, or overly technical phrasing

  • Ensured WCAG 2.1 AA alignment, especially for input fields and error states

Visual & Interaction Patterns

  • Checked alignment with Edward Jones design system / branding

  • Suggested visual affordances for quick actions, buttons, or input states

  • Ensured mobile responsiveness and adaptive layout integrity

Content Strategy & Error Recovery

  • Reviewed system messages for consistency and helpful tone

  • Advised on fallback messaging for when the chatbot couldn’t interpret input

  • Supported efforts to guide users to human support when needed

Quality Assurance & Design Feedback

  • Provided asynchronous feedback through design critiques or written documentation

  • Identified UX debt or areas for iteration before final delivery

  • Acted as a bridge between design system governance and new component creation

Leading Figma AI Integration

My role:

  • Wrote the internal white paper and opportunity framing

  • Ran demos for stakeholders, including GPs (General Partners)

  • Managed internal pilot, set evaluation criteria, and gathered feedback from multiple design teams

  • Published best practices for designers to reference (what Figma AI does well vs. when you should be designing yourself, etc.)

  • Synthesized results and partnered with Dev and DesignOps leaders to recommend long-term tooling decisions

Key goals:

  • Speed up design documentation and discovery

  • Support designers with content scaffolding, not just lorem ipsum

  • Allow designers to create a quick, polished mock-up on the fly

  • Evaluate accessibility and risk (no AI usage in protected client content)

AI-Powered Survey Results

In Edward Jones DesignOps, we wanted to survey the product teams we serve, but our process for consuming those results was completely manual. And the insights that weren’t manual weren’t rich. I took the initiative and came up with a plan to use AI to enrich our results and tell a better story using data. This is still in progress, but it includes:

LLM-Based Summarization

Use an LLM to generate summaries of open-text responses by topic, sentiment, or team. Included prompts like “Summarize the top workflow frustrations mentioned across all responses.”

Text Clustering & Topic Modeling

Use unsupervised machine learning (e.g., K-means, LDA) to group similar responses and identify emerging themes (e.g., “handoff,” “documentation,” “rework”) without being manually biased.

Tools:

  • Python with scikit-learn

  • Power BI AI visualizations

Sentiment Analysis

Assign sentiment scores to open-text responses to detect frustration, excitement, indifference and slice by team or role to find where pain is concentrated

Tools:

  • Python with TextBlob

Response Mapping to AI Opportunities

For each workflow issue, match it to a potential AI solution (e.g., “We spend hours formatting brainstorming results” → “Use Figma AI to affinitize and summarize brainstorming results in FigJam”)

Keyword Extraction & Heatmaps

Extract frequent keywords (e.g., “handoff,” “Figma,” “misalignment”) and visualize their frequency, then create heatmaps by team or workflow phase

Tools:

  • spaCy

  • Power BI

Trendline Detection or Change Over Time

The intention is to send out this survey regularly every trimester. We’ll compare results against previous surveys to find trending pain points or improvements and predict future bottlenecks based on historical data.

My Approach to AI Design

I bring a measured, systems-first approach to AI. That means:

Useful over novel

Avoiding “wow” moments in favor of utility. I prioritize workflows where AI saves time, not just dazzles.

Assistive, not autonomous

AI should enhance, not replace. I design for human-AI collaboration—clear feedback, human override, and graceful failures. Nothing can replace human judgement and creativity, but those humans should have as many tools available to them as possible.

Security-aware

My cybersecurity training helps me spot where AI output might cross ethical or compliance lines (e.g., generating hallucinated facts, leaking sensitive terms).

Data-conscious

When using generative AI, I always consider what data it was trained on, how outputs are evaluated, and how confident a user should feel acting on them.

Projects Demonstrating Applied AI/ML Knowledge

During my MS Data Analytics program, I’ve built and evaluated several machine learning models, gaining a hands-on understanding of how AI systems are trained, tested, and integrated. So I don’t just make user experiences surrounding AI – I offer an in-depth understanding of how the back-end works. Confidence levels and understanding potential errors is beyond relevant to designing the user experience.

Selected examples:

See also, my Data Analysis page.

I’m also working on my Google Cloud Professional Machine Learning Engineer Certification! I’m hoping to have that done in May or June of 2025.