Author: Prasham Shah
May 3, 2026
Studying computer science while building products creates a useful tension. School pushes you toward fundamentals. Building pushes you toward shipping. The best learning happens when the two collide.
At UC San Diego, I have been exposed to systems, machine learning, operating systems, natural language processing, parallel computing, biology, design, and the weird edges where disciplines overlap. The coursework gives me primitives. Projects force me to use them under ambiguity.
The biggest lesson is that models are rarely the whole product. The model matters, but so does retrieval, latency, UX, data quality, evaluation, deployment, and whether the output actually changes a user’s next action. A product can have impressive AI and still fail if it does not fit the workflow.
Another lesson is that taste compounds. The difference between a toy and a useful product is often not one breakthrough. It is a hundred small decisions: what to hide, what to explain, what to automate, when to ask for input, and how to recover when the system is wrong.
I am also learning that the best builders stay close to the edge of what they do not understand. AI, finance, commerce, and infrastructure are all changing quickly, and the only way to develop judgment is to keep building while the ground is moving.
That is the operating mode I want: learn the fundamentals deeply, ship enough to test reality, and keep tightening the loop between theory and use.