AI & Product Engineer
Software Engineering, Product, Data Science
Posted on Sep 10, 2026
We are looking for an exceptional recent graduate or early-career engineer with unusually strong reasoning ability, high agency, and the desire to build something technically and intellectually ambitious. You will work directly with the founders across product engineering, AI systems, data architecture, experimentation, and agentic operations. This is an execution role, but not a mechanical one. You should move fast, question weak assumptions, understand why a system exists, and connect implementation decisions to the larger learning architecture. Priorities will change as evidence improves; you must be ready to pivot without losing rigor. Agentic systems: structured tools, state transitions, evaluations, operational controls, and human-review gates for sensitive decisions. The cultural data model: a growing content genome, the behavioral event spine, and the lineage that supports representation, clustering and forecasting. Falsifiable prediction: cultural and product intuitions turned into explicit hypotheses, tested in product, with predicted compared against observed. Consumer product surfaces: web, mobile, WhatsApp, audio, group ritual, sharing and personalization, locally specific and adaptable across Latin America. You will likely have studied engineering, sciences, mathematics, computer science, or a similarly rigorous discipline at a leading university. Equivalent evidence of exceptional ability is equally valuable. Excellent first-principles reasoning and strong software-engineering fundamentals. Evidence of substantial software built through research, internships, open source, startups or personal projects. At home moving between abstract models and concrete implementation: APIs, databases, data models, tests and user interfaces. Curious about agents, representations, evaluation, experimentation, clustering, causal reasoning and forecasting. An extreme sense of ownership, hands-on by default: you take a system from hypothesis to shipped without waiting to be told. High execution speed, intellectual honesty, coachability, and comfort discarding work when evidence changes the direction. Thoughtful about culture, privacy, and the limits of automated inference. {'kind': 'predict', 'intro': 'There is no application form. There is a system, and you are already inside it. Open the console.', 'steps': ['window.enot is real. Start with enot.help().', "State your hypothesis before the outcome: enot.predict('people'), or any tile, before the ball lands.", 'Make it true twice in a row. Reading the source to shortcut it also counts. That is rather the point.'], 'streak': 2, 'codeWord': 'FALSIFIABLE'}