About Arvor

The root came decades before the fruit.

Arvor was founded in 2024, but it started much earlier: in 1999, inside a search engine. Every tree grows this way — years of invisible root sustaining every fruit that comes later.

This page tells that story: 25+ years making machines find, understand and use knowledge. The discipline changed names several times — search, big data, machine learning, AI. The engineering is the same, and more powerful every year.

Nobody sees the root. Everybody picks the fruit.

Leonardo Dias — founder of Arvor
Leonardo Dias · Founder

The founder

Leonardo Dias

Founder of Arvor. Co-founder and former Chief Data Officer of Semantix, which listed on NASDAQ in 2022. Building search engines and data systems since 1999 — from Lucene to RAG, from Hadoop to agents.

He codes, writes, teaches and hosts podcasts — always around the same obsession: turning raw data into useful knowledge, in production, for real people.

Today he builds Arvor in public: every product, every launch, every hit and miss documented.

The arc

From a search engine to NASDAQ — and what comes next.

  1. 1999 · Catho Online

    From programmer to product strategist

    Thirteen years at Catho, the Brazilian job platform — from programmer to Product Strategy Manager. Owned the company's job and résumé search engines, leading a multidisciplinary team of 13 reporting directly to the presidency.

  2. 2006 · Apache Lucene & Solr

    Search from day one

    Adopted Solr the year it launched. Synonyms, relevance tuning, scale — the same retrieval engineering that powers modern RAG architectures today.

  3. 2010 · Semantix

    Co-founder and Chief Data Officer

    Co-founded Semantix, a Latin American pioneer in big data, search and AI. Figures published by the company itself: around 300 clients, ~10 PB of data under management and 6+ billion transactions per month.

  4. 2022 · NASDAQ: STIX

    One of the first Latin American deep techs on NASDAQ

    In August 2022, Semantix listed on NASDAQ via SPAC with Alpha Capital, under the ticker STIX — a business combination that valued the company at approximately US$1 billion at announcement.

  5. 2024 · Arvor

    Back to the root

    Founded Arvor to build the next layer: AI agents that work and knowledge brains that remember — all built in public.

Credentials

Not a pitch. A track record.

Teaching

Cloudera Best LATAM Instructor

Named Cloudera's best instructor in Latin America in 2013 and 2015. Hadoop Admin/Developer and HBase certifications. Professor in FIAP's Big Data/BI MBA since 2013.

Stage

ApacheCon and the Lucene community

Speaker at ApacheCon USA 2008 and 2009, Lucene Revolution Barcelona 2011 and ApacheCon 2013. Organized the São Paulo Lucene Meetup in 2013.

Production

Enterprise search at scale

Search and recommendation deployments for Magazine Luiza, Itaú, Petrobras and Globo.com (G1) — e-commerce, internet banking, energy and media.

Stack

From the Lucene era to modern retrieval

Hadoop, Spark, Kafka, HBase, MongoDB, NLP, ML — one continuous line from classic search engineering to today's RAG architecture.

Media & books

Builders should also tell the story.

Podcast

Rádio Café

Daily satirical podcast with ~400 episodes. A fully automated pipeline built by the founder himself — script, cover art and assembly.

Podcast

Ressonância Cognitiva

A podcast dedicated to AI: how machines think and what that changes for those who build and work.

Book

IA Para Todos

An introduction to artificial intelligence for non-technical readers (PT). Available on Amazon.

Book

100 Axiomas Para A Vida

Life and work principles distilled into short axioms (PT). Available on Amazon.

Arvor today

Agents that work. Brains that remember.

AI agents in production

AMIGO, Caption AI, Relentless, Brasil CLI and Concierge — agents and products actually running, from app to terminal, from WhatsApp to npm. And Agentic Consulting deploys the same stack into real companies' operations.

See products

Knowledge brains

Memento, AGINUS and Brain Maker — local-first memory and cognitive infrastructure so agents remember, learn and compose knowledge. The natural continuation of 25 years of search engineering.

Explore consulting

Want to build with us?

From agentic consulting to products, everything starts with a conversation.

Contact us