Memory architectures for long-running AI agents
About
Jorge Guberte
Jorge Augusto Guberte is an AI engineer and systems builder based in São Paulo, focused on agent memory, orchestration, generative interfaces, and applied LLM systems.
What I care about
Context engineering, retrieval, and orchestration
Open-source infrastructure for agentic systems
Low-resource language models and language preservation
Elegant systems with strange constraints
Selected experience
Hub Esfera
Co-Founder & Principal AI Engineer
Designing AI products and agent workflows for complex public-sector data, with a strong focus on orchestration, memory, and API-driven systems.
Independent AI systems work
Builder / Consultant
Prototyping agent architectures, retrieval pipelines, and experimental AI interfaces that turn abstract ideas into systems people can actually use.
Earlier software engineering work
Full-stack / Backend
A long foundation in web platforms, APIs, and product engineering that now feeds directly into how I design AI systems.
Stack
Languages
PythonTypeScriptJavaScriptRust
AI Systems
MCPLangGraphCrewAIMastraRAGEvaluation Pipelines
Infra
ModalDockerGCPGitHub ActionsvLLM
Data
PostgreSQLQdrantMemgraphRedisDuckDB
I like systems that feel alive: things that remember, decay, branch, adapt, and become more useful with time.