Francisco Medrano

AI Engineer · Full Stack · Internal Solutions Delivery

I build software with AI and take it all the way to the people who'll use it.

This site is built by a team of AI agents. I decide what goes live.

The AI agent team that builds this siteEight connected agents: Spec writer, Architect, Front end and Back end in parallel, UX/UI, Test, QA and security, and DevOps. A task moves from one to the next. What each one does is explained in the "This site" card further down.1Spec writer2Architect3Front end4Back end5UX/UI6Test7QA & security8DevOps

In progress: story H2 · Home Spec writer

What I do

Three things, and they usually go together. AI is little use if the app breaks or nobody ends up using it.

Agents and AI

I get AI agents working together, with a person in charge.

I split the work into steps, give each agent its own and leave the final say to a human.

  • An agent team that handles personal-data requests in 6 steps, with human review.

    demo
  • LangChain and LangGraph demos, each with its own tests.

    code
  • This site, built by an agent team I direct.

    work log

Try it in the Lab ↓

Full stack

I build the whole application: what you see and what you don't.

Angular on the surface; Java and Python underneath. And tests, so it still works tomorrow.

Internal Solutions Delivery

I turn what's needed into requirements and coordinate the teams through to delivery.

It's what I do today at INDRA: making sure each team knows what's needed, by when, and what it depends on.

View full experience →

Reading about it is fine. Trying it is better ↓

Lab

I built it. Now within your reach.

Two demos from my courses that I'm bringing in here, and a third you're already using: this site.

GDPR Agents

  • multi-agent
  • 6 steps
  • human review

When someone asks to see, correct or delete their personal data, six agent steps prepare the reply. Before it goes out, a person decides.

ChatBot Pro

in progress

  • LangChain
  • Gemini
  • live replies

A chat built with LangChain and Gemini. It answers as it types and remembers what you've said.

This site

  • 8 AI agents
  • Scrum
  • in the open

It's built by a team of eight AI agents along the build flow, each with its own role. I'm the Product Owner: I decide what gets built and review what ships. The graph at the top is that team at work.

  1. Spec writer

    Writes down, on one page, what each part of the site must do.

    specs
  2. Architect

    Decides how it's built and writes down why.

    decisions
  3. Front end

    Builds what you see, true to the approved design.

    in progress
  4. Back end

    Builds what you don't see, in Java.

    in progress
  5. UX/UI

    Checks that what was built matches the approved design: screenshots, motion, light and dark.

    review
  6. Test

    Checks everything works before it's published.

    in progress
  7. QA & security

    Reviews the work with fresh eyes: it starts from scratch, without having seen how it was coded.

    work log
  8. DevOps

    Publishes only when the tests pass and I give the go-ahead.

    in progress

How I work

Four habits I bring to every project. This one included.

  • I work in the open.

    This site's code, plan and work log are public. You can see what's moving, and what's still missing.

    Read the work log →
  • A person decides.

    The agents propose and I review. Same with what I build for others: someone always has the final say.

    See how it works in GDPR Agents ↑
  • I check before I deliver.

    Automated tests and a second review from a clean context before anything goes live.

  • If I can't show it, I say so.

    Much of my work belongs to companies and isn't public. I describe it without client names or internal details, and say where I did it.

    View full experience →

Sound like a fit? Let's talk ↓

From idea to delivery. Shall we start?

If your team has an AI idea to design, build or get into the hands of the people who'll use it, tell me about it.

The contact form is on its way. In the meantime, LinkedIn.