AI Engineering
- RAG pipelines
- MCP servers
- Agent loops & tool use
About me
Full stack engineer, building with AI
Hi! I'm Tomas, a full stack software engineer focused on AI-driven systems. I build the applied side of AI — RAG pipelines, MCP servers, and agent workflows — on top of production web applications built with Next.js, Node.js, and NestJS. Not model training or fine-tuning: the engineering that turns a model into a feature people actually use.
Toolbox
What I work with
LLM Integration
- Azure AI Foundry
- Ollama (local)
- Atlas Vector Search
Backend & APIs
- Node.js
- NestJS
- REST & GraphQL
Frontend & Platform
- Next.js
- React.js
- AWS
What I do
I build AI features end to end and the systems that carry them. That has meant a RAG pipeline with intent classification and semantic search over MongoDB Atlas Vector Search — prototyped locally against Ollama, then moved to Azure AI Foundry for production — plus MCP servers that give models real tools, and hand-written agent loops with the guardrails and fallbacks they need once real traffic arrives. A model call is the easy part; the engineering around it is the work.
Underneath that sits the same full stack discipline: modern interfaces with Next.js and React, robust APIs with Node.js and NestJS, and cloud infrastructure that stays reliable and affordable. I also use AI-assisted development day to day, which changes how much one engineer can reasonably own.
tomydiaz.com serves two purposes: a portfolio of my professional work, and a technical blog where I share the insights and lessons learned along the way.
What I write about
- RAG pipelines — retrieval, chunking, and vector search
- MCP servers and giving models real tools
- Agent loops, tool use, and where they break
- Designing and operating AI features in production
- Full stack architecture underneath all of it
- Code quality and maintainability