Applied AI
Prompting is easy. Production is not.
Applied AI, with your data in-house
Language models running on your own infrastructure and MCP connectors so that AI works with your systems without sending your data to third parties.
01 · What I do
What I do
Models on your infrastructure
Hugging Face models deployed on your servers, with OpenAI, Ollama and Anthropic compatible APIs: the tools that already speak those APIs work against your model.
- LLMs
- Hugging Face
MCP connectors
MCP servers so that assistants such as Claude can query and operate your systems, with authentication in front.
- MCP
AI in development
Legacy code analysis, test generation and documentation assisted by LLMs, always reviewed by a person.
- LLMs
- Testing
02 · HFL in numbers
HFL in numbers
- 3 compatible APIs on the same port (OpenAI, Ollama and Anthropic)
- 29 releases published on PyPI
- 5,600+ automated tests
03 · Own projects
Own projects
HFL
Local Hugging Face model server
Open source (Apache-2.0), with 29 releases on PyPI and 5,600+ tests.
- Python
- FastAPI
current-time-mcp
Hardened remote MCP server
Date and time tools over Streamable HTTP, which can be added as a connector in Claude.
- MCP
About me
Gabriel Galán Pelayo
Senior Backend Developer & Technical Leader
30+ years in IT, 10+ of them building scalable backend systems. Expert in modernizing legacy applications, building high-performance APIs, and leading technical transformations. Specialized in Ruby on Rails, Python, and cloud-native architectures.
- Madrid · on site or remote
- Spanish native · English for international teams
Contact
Let's talk.
Tell me which task you want to automate or which data cannot leave your company, and we'll see which model and architecture fit.