She learned to argue in front of thousands. Then she went looking for something worth arguing for.
Luella Zhang is fourteen. She has stood at Harvard's podium, run a Duke research desk, sat beside patients who didn't speak the language of their own diagnosis, and — somewhere between all of those rooms — decided that the only reasonable next move was to start a cancer lab. This is the long way of saying: she noticed. And then she refused to look away.
In Grade 8, Luella took a job most fourteen-year-olds would never think to ask for — clinical administrative assistant at a UBC Faculty specialist's practice in Vancouver. Officially, she confirmed appointments and chased paperwork. Unofficially, she became the person families looked at when the specialist had already moved to the next room and they still didn't understand what had just happened to them.
Immigrant grandparents. New mothers translating a prognosis into a second language, one word at a time. People she recognized from her own neighborhood. Week after week, she watched the same quiet pattern: the newest, best science almost never reached the people who needed it most. It stopped at the ones who could afford to wait. She started keeping a notebook — the questions no one at the desk had time to answer. That notebook is the first draft of Luella Labs.
Before she was a founder, she was a voice. Five years of competitive debate — Champion at UBC HST, Best Speaker at the BC Provincials, 12th Reply Speaker at the Harvard College WSDC Invitational competing in the Open Division against national senior teams, 3rd Speaker at the Canadian National Impromptu final. Kids twice her age; rooms she wasn't supposed to be old enough for.
Debate did two things to her. It taught her how to hold a room — the pacing, the pause before the hard sentence, the discipline of making complicated things arrive whole in someone else's head. And it taught her the habit that now runs the lab: steelman the other side of your own hypothesis before anyone else can. Luella Labs is run like a debate room. Every scaffold has to survive its strongest counter-argument before it earns a single hour of wet-lab time.
"The patients I sat with couldn't wait fifteen years and three degrees for me to be allowed to try. So I stopped waiting."
From January to May 2026, Luella was a research intern at Duke University's Biomedical Engineering School. Her question was uncharacteristically stubborn for a fourteen-year-old: not "can AI diagnose better than a doctor?" — a question everyone was already asking — but the harder, quieter one underneath it. Why does none of it reach the people I was sitting across from?
She spent months inside AI regulatory sandboxes — the frameworks that let hospitals safely trial new algorithms before a full FDA cycle. The work went on to win a Regional Second Prize at the International AI Innovation Olympiad. And it left her with a thesis she hasn't been able to put down: the real bottleneck in oncology isn't ideas. It's the loop between an idea and a validated readout. Every week you compress that loop is a week someone at that front desk doesn't have to wait. Luella Labs is built to compress it. BRAF-Δ is the first proof.
This site is Luella's notebook, kept in public — milestones and dead ends, funding steps and seminar recordings, the students she is quietly pulling up behind her. It is the lab she went looking for when she was thirteen and couldn't find. She is fourteen now, and she is building it — not for a credential, not for a headline, but because the woman in the waiting room is still waiting. And because the next kid with a question deserves a door that is already open.