David Garcia

AI engineering

I build AI agents around your product knowledge and workflows.

AI agent consulting, RAG, MCP integration, and evaluation. From a focused review to implementation.

What I can help with

  • Agent workflows

    Build assistants that use your product knowledge and call the tools needed for an agreed task. Define the workflow, access, review steps, and recovery when a step fails.

  • RAG development

    Build retrieval-augmented generation around your documentation and product knowledge. Connect sources, improve retrieval, handle updates and access rules, and test answers against representative questions.

  • MCP integration

    Make product knowledge and APIs available through the Model Context Protocol. Design tools and resources, integrate authentication and permissions, and test the interaction in the clients your team uses.

  • Agent evaluation

    Find why an agent gives the wrong answer or fails a task. Inspect retrieval and tool calls, build repeatable evaluations, and check changes against quality, latency, and cost requirements.

David Garcia

Who I am

I’m David Garcia, a software engineer and founder of Biel.ai and TechDocs Studio.

At Biel, I work on turning technical documentation into source-linked answers and context for AI agents: retrieval, integrations, evaluation, and the engineering needed to run the product.

Explore Biel.aiRAG APIMCP integration

Working together

We start with one workflow: who uses it, what it needs to do, and what a successful result looks like. I review your existing system and agree the deliverables and project price with you.

Implementation happens in your repositories, with working changes and evaluations your team can review. The handover includes the source, checks, and instructions to run and improve it.

The stack follows the problem. We can build on your current tools or assess whether an existing product covers the need.

Technologies

  • LangChain and LangGraph for agent workflows.
  • LangSmith for tracing and evaluation.
  • OpenAI and Anthropic for model integration.
  • Typesense, Qdrant, and Pinecone for retrieval.
  • Cohere for reranking.
  • Python and FastAPI for APIs and MCP services.

Let’s discuss your project.

Discuss a project