Artificial Specialized Intelligence
Production AI systems rely on expensive, slow frontier LLMs for narrow, repeatable tasks where general-purpose intelligence is unnecessary, limiting speed, cost, and accuracy.
ZeroEntropy trains small, task-specific models (e.g., for reranking, embeddings, classification, routing, query rewriting, and context compression) to replace frontier LLMs in AI pipelines.
Developers and AI product teams building production AI systems like copilots, agents, and search applications.
cofounder & CEO @ ZeroEntropy (YC W25) - moroccan founder with 2 masters in applied maths from École Polytechnique and UC Berkeley
Cofounder & CTO @ ZeroEntropy (YC W25) - My background is in theoretical mathematics and computer science. Dropped out of CMU to pursue startups. Worked on low-level C/C++/Assembly/GPU Code. CTO and/or main dev for 5 different startups. Created stat-arb algorithms and audited blockchains for bug bounties and hedge fund due diligence. Built the AI @ myko.ai, manifestapp.xyz, magibook.co



