AP Research: Study of LLM Memory in Agentic Tasks
arrow_outwardConducted a 90-trial quantitative experimental study comparing Stateless, Recurrent, and Retrieval-Augmented Generation memory architectures for LLM agents in a full Catan simulation.
I'm Samhith, a developer in high school who's interested in Computer Architecture, embedded systems, and machine learning. Check out my projects!
Conducted a 90-trial quantitative experimental study comparing Stateless, Recurrent, and Retrieval-Augmented Generation memory architectures for LLM agents in a full Catan simulation.
Engineered real-time flight control logic for a custom Vertical Take-Off and Landing aircraft, mitigating the 'Sim-to-Real' gap in unstable aerodynamic environments.
A database-free Content Management System optimized for minimal operational overhead, utilizing an aggressive in-memory mapping cache instead of SQL.