I build real-time graphics systems and work on AI tooling, mostly focused on practical applications and internal tools.

• Built a local LangChain-based agent running Qwen3:14B through LM Studio
• Integrated custom tools using the Model Context Protocol (MCP)
• Developed a SpaceTraders MCP server for autonomous interaction with the game state
• Set up a local agent interface using LangGraph’s UI
• Working on privacy-focused assistants for internal workflows and development use

Over the past few years, I’ve worked on real-time visual systems using tools like Ventuz and Unreal Engine.

Projects have included esports broadcast systems for Riot Games, as well as installations for organizations like Yale and the NHL. A lot of that work involved live data, rendering pipelines, and integrating real-time inputs into visual outputs.

More recently, I’ve been experimenting with combining those systems with generative models.

• LLM-based agent systems that can actually be used in day-to-day workflows
• Building tools that run locally and don’t rely on external APIs
• iOS and Swift development, especially for real-time or generative media
• Simulation systems, game logic, and agent-driven environments

• Portfolio: https://portfolio.hugovalverde.com/
• Blog: https://blog.hugovalverde.com/
• Instagram: https://www.instagram.com/thehugovalverde/
• YouTube: https://www.youtube.com/channel/UCp86sevt4raRJB5up5a3nxg
• LinkedIn: https://www.linkedin.com/in/thehugovalverde/

• Python
• C#
• Rust
• Unreal Engine
• Ventuz
• DaVinci Resolve