About
The kind of PM who asks why before what.
I'm Alex Cruz-Valencia. I grew up in Salinas, California, I'm a first-generation college student and QuestBridge Scholar, and I study Computer Science and Human-Centered Design at Dartmouth.
Back home, I was a member at Digital NEST, a nonprofit in Salinas that gives young people a path into tech, and later an instructor there. Last winter I went back and proposed my own research study on why new members stall out, and one of my recommendations is now how the org talks to its members. That's still my favorite kind of work: knowing a group of people well enough to see what's actually in their way.
Today I'm the product manager for Evergreen, Dartmouth's student-built AI wellness app, backed by $5M in research funding and in testing with about 200 students. I joined as a project assistant, got promoted to PM, and led its design team through the first build. Now I lead design and engineering, and I own the parts of the AI that decide whether students trust it: what Evie can do on their behalf, and what it remembers about them.
My work sits between building and understanding. I can read a schema and I can run a user interview, and I think both belong in the room when a product decision gets made. This summer at EchoStar that meant shipping a self-serve dashboard on Snowflake and deleting a feature I'd built once I checked my assumption with the people it was for. At Youphoria it meant arguing for a social feature, Groups, and owning it all the way to launch.
The through-line is a pull toward products that help people live a little better, and a habit of testing my own assumptions before I trust them. It's why my side project, Puente, is still in validation instead of in the App Store: I'd rather find out it's the wrong idea in a conversation than after a build.
I speak English and Spanish. Outside product work, I box, play poker, sew and tailor my own clothes, and build custom keyboards.
How I work
Lessons I paid for- 01
When I can't win the decision, I win a way to learn it.
Leadership rejected my permissions recommendation twice. I stopped arguing and got them to commit to an A/B test instead.
The full story → - 02
Constraints are the spec.
Security rules, data sensitivity, and what execs actually wanted carved an 8-page dashboard down to the 4 it should have been.
The full story → - 03
Find the number that changes the question.
17 people joined, 5 finished onboarding. That ratio turned "market better" into "stop losing people at the door."
The full story → - 04
Comply with the directive, protect the work.
Told to stop interviews for focus groups, I kept a safety net running. The focus groups never happened. The research survived.
The full story →
Toolkit
What I reach forResearch
- Interviews
- Surveys
- Journey mapping
- Usability testing
- Synthesis
Product
- Figma
- Jira
- Miro
- Notion
- Experiment design
Data
- SQL
- Snowflake
- Streamlit
- Python
AI & building
- LLM tool calling
- Memory design
- Claude Code
- Cursor
- Codex
- Gemini Gems
- Next.js
- TypeScript
On campus
- Officer, Dartmouth Society of Computer Science Students
- Dartmouth Tech Consulting Group (formerly UX/UI designer)
- QuestBridge Scholar
- SHPE · ALPFA · ColorStack