Good ideas survive contact with the real world.

Competitive questions arrive inside live deals, where static battlecards are usually too broad or already out of date. I built Intellibot to generate account-specific comparisons and bottom-up TCO analyses from evidence and assumptions that remain open to inspection. The public demo uses dummy data; the working version remains internal.

AI serving costs rarely appear as a standalone line item. I built an interactive 10-year view of 56 public software companies for a pulse on how gross margin, free cash flow, and pricing power are trending in the AI era. It does not attribute margin changes to AI; it makes the assumptions visible for outsiders to observe.

I built 100 small drawing systems to explore a larger idea: math does not remove ambiguity; it gives ambiguity structure. Each study is generated from a formula, drawn as inline SVG, and disturbed by one spring value. Math produces possibility, but we get to decide which possibilities are worth pursuing.

Many jobs of 2036 do not exist today. For Autodesk's launch to the youth market, I argued that the durable advantage is not fluency in one tool, but the ability to learn, build, and lead as tools change. That argument became an ISTE21 virtual keynote and an essay titled “Machines Are Getting Smarter, but the Most Important Intelligence Isn’t Artificial”.

Each essay begins with something that happened: a product shipped, a customer conversation, a factory, a classroom, a place, or a decision whose consequences are still unfolding.

Nailed It! follows people through AI workflows they have already made useful—not hypothetical demos. Each interview begins with a real problem, shows what changed, and isolates what another person could borrow. The measure is useful work, not more AI.

I first encountered UNICEF's lifesaving work after the 2008 Sichuan earthquake in my home province. That experience grew into work on a national program that equipped more than 800 chapters and 30,000 young people to organize locally inside a global institution. The challenge was to provide enough structure to protect the mission without crowding out local judgment.

A cause can matter deeply and still leave us unsure where to begin. Through DoSomething.org's Youth Advisory Council and my local high school club, I helped shape calls to action that were specific enough to start, meaningful enough to matter, and honest about what one participant could change. Campaigns spanned entrepreneurship, STEM access, digital literacy, and prom access.

Free does not automatically mean accessible. As a youth advisor, I helped shape the messaging, launch, and PR for Official SAT Practice on Khan Academy around the barriers students actually faced: credibility, mobile access, limited counseling, and no commercial preparation option. Within a year, the service reached 1.4 million unique users—four times the annual population of commercial test-prep classes.

Inside the U.S. Department of Education, I reviewed $72 million in higher-education grants and renewals reaching institutions that served 65% of U.S. enrollment. Rather than compress risk into one score, each recommendation separated what the record established, what remained uncertain, how much public exposure sat behind it, and what evidence should change the decision. The goal was reasoning the next reviewer could inspect.

My cofounder Mark Pavlyukovskyy and I built Mochi to bring the Montessori method to STEM. Our goal was to encourage more creators, not just consumers, of technology. Today, Mochi has reached 1000s of homes and schools in 63 countries (and counting)!

Abrupt mandates for nationwide school closures can be daunting. With support from the Gates Foundation's US program, we equipped 10,000s of K-12 educators and school leaders with the tools, tips, and resources to navigate the transition. Within 90 days, I worked with a team of 5 educators to package and distribute 100 resources, then reported on improvement stories through summer PD and back-to-school.

Tinkercad already made 3D design approachable for individuals. I led the global relaunch that repositioned it as Autodesk's K–12 front door, aligning the brand, website, acquisition path, and the workflows teachers, students, and administrators needed. The launch added 20 million learners to a community of more than 100 million and drove more than 100% year-over-year growth.

Customers were already pairing Azure and Datadog. My role was to turn that customer pull into an enterprise story across marketplace distribution, Azure billing, telemetry, integrations, and joint support. Making the experience native helped teams build, run, and migrate applications with less friction while keeping migration risk and cloud ROI visible.

To go upmarket and reach enterprise CIOs, CTOs, and CISOs, I interviewed seven technology leaders about the architectural decisions, operating constraints, and behavior changes behind resilient cloud systems. A shared interview framework made the stories comparable without flattening each leader's context. The series positioned observability as a leadership discipline tied to cloud ROI, organizational agility, and customer experience.

I built a TikTok series from zero to learn short-form storytelling and test whether machine-learning ideas could travel into everyday life. Each 90–120 second episode was scripted, recorded, and edited in TikTok. Within six months, the channel reached 14,000 followers and more than 100,000 engagements across the main series and viewer Q&A.