Strategic LLM Agent with RAG & Minimax

Explored how LLM-based agents can operate in a competitive environment where decisions depend on incomplete information, changing state, and strategic look-ahead.

Built and experimented with a competitive LLM-based agent combining retrieval-augmented context with minimax-style strategic reasoning.

The project provided a practical environment for studying how language-model reasoning can be combined with classical search and decision-making techniques.