I speak about the systems that make AI trustworthy, data products understandable, and teams more effective.
Whether it’s a keynote, panel, or hands-on workshop, I bring real stories from designing AI-powered data infrastructure at Microsoft — not theory, not hype, not slides full of acronyms.
Talks
Your AI Isn’t Confident — It’s Ambiguous
Why most AI failures are semantic design bugs, not model problems
Modern AI systems appear intelligent until they’re asked to operate inside real organizational contexts. What breaks first isn’t reasoning — it’s meaning. This talk reframes hallucinations and trust failures as semantic design problems, and gives teams concrete patterns for making meaning explicit: semantic contracts, assumption surfacing, confidence signaling, and interaction models that treat uncertainty as actionable.
Best for: AI/ML conferences, platform architecture forums, senior IC audiences
Judgment Is the Last Mile
Designing AI systems for decisions humans are accountable for
AI doesn’t own outcomes — humans do. Yet most AI systems are designed as if producing an answer is the finish line. This session focuses on the decision layer: decision checkpoints, structured options instead of raw output, traceable assumptions, and escalation paths. We’ll treat governance as a UX problem.
You’ll walk away with: A framework for designing decision points into AI workflows, concrete patterns for preserving human accountability without slowing teams down, and a new way to evaluate whether your AI system is actually helping people decide or just generating output.
Best for: Leadership forums, Responsible AI events, engineering management conferences
AI Made Us Faster. So Why Are We Slower?
The hidden coordination tax of AI at scale
AI increases output, but many organizations feel more chaotic, not more aligned. This talk exposes the hidden coordination tax: how speed without clarity amplifies noise, decision fatigue, and semantic drift. You’ll leave with patterns that restore leverage: constrained generation, embedded decision rights, and knowledge governance that prevents entropy.
Best for: Flagship conferences, strategy offsites, closing keynotes
Stop Building Assistants. Start Building Mentors.
What teaching builders reveals about the AI interaction patterns that actually change behavior
Most AI systems optimize for “answer delivery.” But humans don’t become better through answers alone. After years of mentoring builders and designing enterprise AI systems, I’ve become convinced that the most powerful AI UX pattern isn’t “assistant” — it’s mentor. This talk shares concrete interaction patterns for building AI that improves user skill over time.
Best for: Hackathon keynotes, community events, podcasts, TED-style formats
Clarity Is Infrastructure (Workshop)
Designing AI (and teams) that think before they speak
A hands-on workshop where participants apply content design principles to real data and AI products. From naming conventions to error message writing to semantic model design, participants leave with a framework for evaluating and improving the clarity of any data product.
Format: 90 minutes, interactive, up to 30 participants
Recent events
| Event | Role | Location |
|---|---|---|
| ODSC East | Speaker, Open Data Science Conference | Boston |
| Microsoft × Red Bull Startup Innovation Labs | Panelist, AI and startup innovation (alongside AMD, WHOOP, Terawatt Energy, Northeastern) | Microsoft NERD Center, Boston |
| Microsoft Design Week | Presenter | Microsoft (internal) |
| Boston University × Data Science AI Hackathon | Speaker & mentor | Boston University |
| PyLadies Boston | Workshop leader, How to Use GitHub, Git Control, and VS Code | Boston |
| Massachusetts AI Hub Hackathon | Mentor & author, wrote blog article + facilitation script that made the event repeatable | Boston |
| Boston Launch Event | Microsoft Design + AI representative | Boston |
| Microsoft Global Hackathon (2022–2025) | Organizer / mentor / judge, 3× executive awards, multi-year | Virtual |
| BU DS+X Hackathon | Mentor & judge | Boston University |
Formats I do
- Keynote (30–45 min)
- Conference talk (20–30 min)
- Panel (moderated or participant)
- Workshop (60–90 min, interactive, up to 30 people)
- Podcast / interview
- Corporate offsite (custom topic)
“Bad data can lead to bad analyses, and bad analyses can lead to bad decisions.”
— The dbt Viewpoint, dbt Labs (source)
Logistics
- Based in Boston, MA — available for in-person events in the Northeast and virtual worldwide
- Comfortable with hybrid formats
- I bring my own slides and can adapt to your event’s visual template
- A/V needs: Laptop connection (HDMI/USB-C), lapel mic preferred, confidence monitor if available
- Lead time: 4+ weeks preferred for new talks; 2 weeks minimum for existing talks
Writing
Many of these talks started as articles. Read the thinking behind the talks on Medium or on my Articles page.
Book me
Ready to have me at your event?
Book 30 min with Jen → · Email me →
I’ll respond within 2 business days with availability and any questions about format, audience, or topic customization.