Andrej Karpathy
Andrej Karpathy is a prominent AI researcher and entrepreneur, known for his work at Tesla, OpenAI, and as the founder of Magical. He is a key figure in the development of modern deep learning frameworks and large language model (LLM) applications.
Recent Developments & Methodologies
Prompting 2.0
In August 2026, Karpathy introduced a new prompting paradigm focusing on leveraging unstructured input for advanced LLM comprehension, marking a shift from traditional structured prompt engineering.
- Core Concept: “Prompting 2.0” utilizes unstructured data inputs to enhance model understanding and output quality, moving beyond rigid template-based approaches.
- Impact: Early demonstrations suggest significant improvements in model reasoning and adaptability.
Karpathy Loop Engineering
The “Karpathy Loop” refers to a structured, iterative workflow design for AI agents, enabling autonomous optimization and complex task execution beyond single-prompt instructions.
- Concept: Focuses on designing iterative loops where AI agents can autonomously perform, evaluate, and improve tasks.
- Application: Demonstrated to significantly enhance the performance of coding agents (e.g., Claude Code) by moving away from static instructions to dynamic, self-correcting workflows.
- Reference: Karpathy Loop Engineering: AI Agent Autonomous Optimization for Development
- Source: Karpathy Loop Engineering: AI Agent Autonomous Optimization for Development