Entrepreneurship and R&D

April 2026↑

Why True Software Developers Won't Be Threatened by AI↑

Thesis: A software developer's core competence was never writing code itself, but acting as a builder who creates things from 0 to 1 — deciding what to build, sequencing the work, making opinionated technical design decisions, and caring about code aesthetics and architecture. Since AI primarily automates code writing rather than these higher-order judgment calls, true developers are not threatened by it; instead, AI lets them delegate the mechanical parts of coding (like a chef delegating chopping to assistants) and focus on the core creative and architectural work, unlocking a 5x–10x productivity gain [1].

The Applied Scientist's Identity Crisis: Why We're More Than Model Builders↑

Thesis: Applied scientists are misunderstood as merely "model builders," but their true mission — echoing Peter Norvig's three eras of AI (algorithm-centered, data-centered, human-centered) — is to research the unknown: defining meaningful research questions, forming and rigorously testing hypotheses, and building frameworks that answer previously unanswerable questions. In the LLM era, where pre-trained models handle much of the technical heavy lifting, this distinction sharpens: training models via established methods is really ML engineering, while applied science is about venturing into uncharted territory to create new knowledge. Applied scientists are defined not by the models they build, but by the unknowns they illuminate [2].