Ramón Hernández

– – Built a desktop application for working with both local and external AI models,

including systems such as ChatGPT and Claude.

– Integrated and optimized local LLM execution using llama.cpp for efficient on-device

inference.

– Designed and implemented autonomous AI agents capable of executing complex

multi-step workflows.

– Developed document analysis and processing systems for extracting, structuring,

and utilizing information from unstructured data.

– Contributed to the development of an AI-powered IDE inspired by Cursor, focused

on agent-assisted development workflows.

– Implemented context chunking strategies to manage limited context windows and

improve long-form AI interactions.

– Designed memory management systems to optimize RAM usage during local

model execution and multi-agent workflows.

– Integrated local and cloud-based LLMs into a unified architecture for flexible AI-

driven task execution.

– Designed orchestration layers for coordinating AI agents across different tools,

models, and execution environments.

– Diagram generation from data using React flows.

– Design of a synchronization system via sockets in Node.js with different types of devices. – Platform for the hotels and restaurants sector to improve menu management in buffets. – Control panel for end customers. – Control panel for distributors. – Control panel for wholesale companies. – Synchronization at the time zone level. – Nodejs with decorators, Nextjs + HeroUI, tailwind, nginx.

– Improve UN documents Chatbot with python and tensorflow. – Create a dashboard to train the neural network online based on response scoring. – Custom D3 Diagrams (Sankey, trees, graphs). – Dashboard to manage teams (roles, permissions). – Backend NodeJS and NextJS.