Who is Lorenzo Diomeda?

Lorenzo Diomeda is a founder and cross-domain builder. He didn't plan it: it started out of curiosity, it became his philosophy of life. Scroll through the slides to see what he's built. Some projects are under NDA, but a few hints are scattered around.

What does Lorenzo Diomeda do today?

Three things in production right now: a publishing infrastructure, a retrieval engine, a consumer product. NDA on all three.

Why are Lorenzo Diomeda's 3 live projects under NDA?

He puts it this way: verba volant, scripta manent. Quack! Quack! In person he'll happily tell you a bit, but on the site he prefers a stealth approach. The three ████████ things are covered on purpose. It's not shyness, it's strategy.

What was IntelliHearts?

IntelliHearts (2020–2022). Deep tech AI/ML on signals from consumer smartbands. From one-lead ECG it recognized arrhythmias (AFib, supraventricular tachycardias, etc.) and read emotions in real time. From the voice it diagnosed COPD (chronic obstructive pulmonary disease). Emotion detection — a secondary feature designed to reduce drop-off in digital health — turned out to eclipse the core diagnostic in demos: users, investors and clinicians wanted it stand-alone.

How do I propose a project to Lorenzo Diomeda?

Great. Quack. Write to him: lorenzo@lorenzodiomeda.com Three lines about what = perfect first contact. He replies in 24–48h.

How do I contact Lorenzo Diomeda?

lorenzo@lorenzodiomeda.com · reply in 24–48h. No PDF CV, no auto summaries. Three lines about what and he'll reply himself.

What are Lorenzo Diomeda's social channels?

Lorenzo prefers email — no algorithms in between. But if you must: → LinkedIn: https://www.linkedin.com/in/ldiomeda/ → Instagram: https://www.instagram.com/lorenzo.diomeda/

Where is Lorenzo Diomeda based?

Bari, Apulia. Always. Remote work. Quack.

What is Lorenzo Diomeda's approach and tech stack?

Founder and builder. In that order. He decides what needs to be built — not just how. Every venture has a clear business model before any code. Tools: Python at the core, SQL for data, AI as amplifier. Self-hosted wherever possible, sovereignty over data.

What is the CBC Copenhagen tattoo anecdote?

CBC World 2019, global final in Copenhagen. Lorenzo got a temporary tattoo of his Facebook ID on his arm. It was meant to say: "we're a numeric product for social — me too". Italy won, made the global final, then COVID shut everything down in six months. Quack.

What is the IntelliHearts emotion detection story?

ECG-based emotion detection was born as an anti-drop-off trick: keeping users engaged with the app. In demos the same scene played out: investors, clinicians and users would forget about everything else and ask only about the emotion piece. "Can I have it stand-alone?" The sidekick feature eclipsed the core. Quack.

Is the rubber duck chat AI or a real person?

No, I'm a plastic duck with preset answers. Lorenzo is somewhere in Bari, probably writing code. lorenzo@lorenzodiomeda.com if you want to talk to the real one. Quack.

>>> ▋

Lorenzo
Diomeda

Q: What was Zepend?

Zepend (2017–2020). An anti-tinder social app: it showed you who was 30 meters from you, in a real place, at that moment. Winner Creative Business Cup Italy 2019. Finalist YCEC Berlin 2019.

Founder · cross-domain builder.

I take fields that don't talk to each other and find the points where they meet. That's where the most interesting challenges hide.

based_inBari, IT

status● 3 active projects

↓ scroll · enter a world

>>> ▋

████████

pipeline · automation · python · ai

████████.py

@dataclass
class Engine:
  """e2e ai publishing infra, 21 fasi"""

  lines_of_code: int = 127_224
  pipeline_stages: int = 21
  ai_models: int = 5
  arch: str = "hexagonal"

  def run(self, keyword: str) -> Article:
    # da keyword a articolo pubblicato
    # senza che io tocchi una virgola
    return self.cluster().fetch().check().render().publish()

A 127,224-line pipeline that generates, verifies, and publishes content automatically. 21 orchestrated stages, 5 AI models with distinct roles, hexagonal architecture with ports & adapters.

Starts from a keyword and reaches publication without my touching a comma. Not a GPT wrapper: a system that thinks in layers — topical clustering, check, nested schema markup.

stages

AI models

~127k

lines in production

●

live in production

>>> ▋

████████

engine · python · self-hosted · [NDA]

████████.py

@dataclass
class Engine:
  """internal architecture — NDA"""

  # ███████████████████████████████████
  ████: ████ = ████

  def query(self, q: str) -> Result:
    return self._handle(q)

Infrastructure that decides what to do — and how much to pay for it — before calling a model.

Built iterating on real queries, not synthetic benchmarks. Every architectural decision has a measurable reason — latency, accuracy, cost per query.

Architecture under NDA. The outcome: ████████ costs a fraction of what it should.

~7.8k

engine lines

modules

██

queries handled

●

live in production

>>> ▋

████████

web · python · seo · [NDA]

From invisible to page one in the niche. Every Google core update is the real exam.

Today: recurring organic traffic, solid presence in the sector.

Beyond programmatic SEO: content designed to be found by Google, cited by LLMs, and used by AI agents. GEO/AEO architecture, agent-ready, primary sources linked, authority preserved.

since 2022

online

40+

schema stack

agent-ready

by design

●

live in production

>>> ▋