01About
Giulliano Soares
Head of Data & AI · since 2001
25 years of building where data, software and product meet. In order, from 2001 to today.
I lead data and AI as a player-coach: most of the time on business impact and decision support, part on leading the team and the rest on platform engineering. Warehouse, governance, agents and AI adoption, from the client's data to the indicator the board uses.
02Results
What changed, in numbers.
Measured in the operations I led. Estimates are marked as such.
45 → 14
days for a new client to go live, before and after
14 → 5
days to close the monthly reports, before and after
approx. 90%
less execution time on the automations
100%
AI adoption across the company, through workshops and one-to-one sessions
100+
non-profit organisations using the products
4,000
daily visits to the agency's own content operation
150M+
people reached worldwide by Impostômetro since launch
2,500
leads per month
03Cases
Problems I solved as a leader.
2024–present
Head of Data & AI · Liquid AI
- Problem
- A new client took more than a month to go live, the monthly report took two weeks, every area read the same data its own way, and AI stayed with those who already knew how to use it.
- What I did
- I led a team of three engineers as a player-coach. I built the Medallion warehouse on BigQuery in Google Cloud, with dbt on the indicators, contracts, dictionaries and quality checks on every pipeline against metric drift, and lineage from the client's file to the indicator. I replaced the BI tool with an internal AI application that queries the warehouse in natural language, automated contract and document analysis, lead generation and data validation, and took AI to the whole company through workshops and one-to-one sessions.
- Result
- A new client now goes live in weeks, the monthly report closes within the week, the automations run in a fraction of the time, and the whole company uses AI in its work.
45 → 14
days for a new client to go live, before and after
14 → 5
days to close the monthly reports, before and after
approx. 90%
less execution time on the automations
100%
AI adoption across the company, through workshops and one-to-one sessions
2020–2024
Product Ops Manager · RiskNow.ai
- Problem
- The portfolio-monitoring product reached every client on a single date, and deployment, maintenance and support ran outside the product.
- What I did
- I led the product and the deployment and maintenance operations. I designed the Data Lake, the Warehouse, the Lakehouse and the ELT that fed the product, and the staged rollout models. I owned the process across sales, admin, finance, legal and customer relationship, from discovery to delivery.
- Result
- Releases started arriving in stages, and what failed in deployment and support became roadmap.
2019–2020
Product Manager · Trackmob
- Problem
- Civil-society organisations depended on the same products to acquire donors, keep the relationship and analyse the result.
- What I did
- I was accountable for the acquisition, relationship and analysis products, the BI and AI line, and ran them as an operation, with discovery inside each client's own operation.
- Result
- The products served Greenpeace, Doctors Without Borders, the ICRC, ActionAid and more than a hundred other organisations.
100+
non-profit organisations using the products
2014–2018
Co-founder and Planning Director · Nossa Causa
- Problem
- A new agency needed its whole operation: marketing and sales, admin, legal and finance, development, content and media.
- What I did
- I co-founded the agency and structured every process of the operation, from discovery to delivery, with Scrum, Kanban and GTD.
- Result
- The content operation started producing traffic and leads every day, and the agency ranked among the best performance agencies in Resultados Digitais' Espada Ninja, on its own account.
4,000
daily visits to the agency's own content operation
2,500
leads per month
1,000+
articles published
2nd
place in the Espada Ninja award from Resultados Digitais, second-best performance agency for its own account
2012–2013
Digital Planning Director · IBPT
- Problem
- How much tax each person pays sat in public data that almost nobody could read.
- What I did
- I ran Impostômetro, Lupa no Imposto and Empresômetro, public products that publish tax data in real time.
- Result
- Impostômetro has reached people all over the world since launch, and Lupa no Imposto helped drive the federal law that puts the tax on the receipt.
150M+
people reached worldwide by Impostômetro since launch
Brazilian federal law requiring taxes to be shown on every receipt, which the Lupa no Imposto project helped drive
04Trajectory
Every position
Every position, from the latest to the first, with the capabilities each one called for.
to PresentSão Paulo, Brazil (remote)
Head of Data & AI
Liquid AI
I lead the data and AI function at Liquid AI: governance of the data pipelines, the security and privacy framework around them, a company-wide AI-first culture, the intelligent agents that automate critical workflows, and the DDC Framework that governs how AI-assisted work gets delivered. The product context is the Brazilian real-estate market, with product-led growth, data-driven decisions and AI-led features.
- Direct the business and the product from the data, building the metrics and indicators the market runs on for the business, finance, sales, product and customer teams.
- Built the Medallion pipeline: a model validates structure and business rules on arrival, then drives Bronze, Silver and Gold with a human in the loop. Unstructured input from many sources, enrichment against data bureaus, and lineage from the client's file to the indicator it feeds.
- Mapped the company's obligations under LGPD and GDPR and structured the security and privacy framework on top of that mapping, so each control exists because a specific rule requires it.
- Put governance on the data pipelines, with ownership, contracts and quality checks, in place of ad-hoc extracts nobody owned.
- Drove AI adoption across the whole company through mentoring and training on LLMs for productivity, product and AI-assisted development.
- Built the intelligent agents that automate critical workflows and scale product and development delivery.
- Wrote the DDC Framework: four layers of doctrine (business, product, engineering and operations) that a coding agent reads before it writes, with 18 imperative rules, 8 modelling contracts and decision records, so AI-assisted delivery is governed rather than supervised.
Capabilities
- Vibe Coding and AI-Based Products
- Data Architecture and Governance
- AI Culture and Operations
to
Founder, Product & Technology Lead
Yala AI
I conceived and shipped two products on Google Sheets on my own, from discovery through to delivery: I researched the market, designed the business model and the subscription plans, worked out cost and price, and only then launched. Yala orchestrates agents, memory and variables inside the cells. Mail Deliver sends e-mail from the same spreadsheet, with templates and open tracking in Analytics.
- Took both from hypothesis to a paid release on the Google Workspace Marketplace, directing agents instead of hiring a specialist for each layer.
- Designed the pricing and wired Stripe myself, which is the part of a solo product that decides whether it is a product or a demo.
Capabilities
- Vibe Coding and AI-Based Products
to
Product Ops Manager
RiskNow.ai
I led the portfolio-monitoring product and every deployment and maintenance operation around it, designing the rollout models, the Data Lake, Warehouse and Lakehouse architecture, and the continuous improvement of the ELT and support processes.
- Designed the Data Lake, Warehouse and Lakehouse architecture behind the product.
- Designed rollout models so a release lands in stages instead of on one date for everyone.
- Owned the process across every area the product touched: commercial, administrative, financial, legal, customer relationship, and discovery through delivery, run with Scrum, Kanban and GTD.
- Ran deployment, maintenance and support as part of the product, feeding their failure modes back into the roadmap.
Capabilities
- Data Architecture and Governance
- Vibe Coding and AI-Based Products
to Curitiba, Brazil
Product Manager
Trackmob
I was accountable for the acquisition, engagement and analytics products, the BI and AI line, used by Greenpeace, Doctors Without Borders, the ICRC, ActionAid and 100+ other non-profit organisations.
Numbers
100+
non-profit organisations using the products
Capabilities
- Vibe Coding and AI-Based Products
to
Founder and Product Manager
OnlySheets
A playground of my own for turning spreadsheets into working software. I structure and validate the models that automate business, product, marketing and sales work for startups, then keep the ones that survive contact with a real workflow.
Capabilities
- Vibe Coding and AI-Based Products
to Curitiba, Brazil
Marketing Director
Trackmob
I owned brand positioning, communication and marketing automation, qualifying hundreds of leads a month. The work also covered strategic partnerships and inbound and outbound automation, ran with a lean marketing and sales team, and consolidated the company's authority in the Brazilian non-profit sector.
Capabilities
- Vibe Coding and AI-Based Products
to
Founder and Product Manager
YouSendr
I structured the business model and ran discovery, branding, the design system, documentation and prototypes.
Capabilities
- Vibe Coding and AI-Based Products
to
Co-founder and Planning Director
Nossa Causa
I co-founded Nossa Causa, a social-transformation agency, and structured its creation, media and sales processes. They ran end to end and covered every area a founder ends up owning: creation, development, media, sales, administrative, financial, legal and customer relationship, across both the institute and the agency, today called BeCause.
Numbers
4,000
daily visits to the agency's own content operation
2,500
leads per month
1,000+
articles published
2nd
place in the Espada Ninja award from Resultados Digitais, second-best performance agency for its own account
Capabilities
- Vibe Coding and AI-Based Products
to Curitiba, Brazil
Teacher of Discipline, Focus and Habits
A Grande Escola
I taught the methods and tools that helped hundreds of people build habits, reach goals and act on ideas. The techniques were about setting the rules of your own life and building habits that carry you anywhere.
Capabilities
- AI Culture and Operations
to Curitiba, Brazil
Digital Planning Director
Getz Comunicação
I built the agency's digital department from scratch: planning, creative, media and development. The creative team covered design and copy, the media team was digital, and the department won the work for clients such as Gazeta do Povo, Grupo Positivo, Neodent, Super Festval, Atlantic Energias and Paraná Banco.
Capabilities
- Vibe Coding and AI-Based Products
to Curitiba, Brazil
Digital Planning Director
IBPT
I ran the Impostômetro, Lupa no Imposto and Empresômetro websites, which publish tax data in real time. The remit was institutional communication and the creation and upkeep of every public-interest site the institute ran, Feirão do Imposto and Amiga dos Sonhos included, publishing tax-collection and business-formation data.
Numbers
150M+
people reached worldwide by Impostômetro since launch
Brazilian federal law requiring taxes to be shown on every receipt, which the Lupa no Imposto project helped drive
Capabilities
- Data Architecture and Governance
- Vibe Coding and AI-Based Products
to
Project Manager
Salsa:Mobion
I managed delivery with agile methods for clients such as Gazeta do Povo, Paraná Banco and Volvo. The client list also included Opet, Oxford Porcelanas and Super Festval.
Capabilities
- Vibe Coding and AI-Based Products
to
Full-Stack Developer
Freelance
I worked as a full-stack developer specialising in e-commerce, building stores for the Curitiba market. It ran for three years, between the agency I had founded and my return to a salaried role.
Capabilities
- Vibe Coding and AI-Based Products
to
Founder
Agência da Internet
I founded my own agency, co-founded the first Impostômetro and launched the first SóCarrão portal. I ran the business end to end: finance, administration and operations, while working as its graphic and web designer.
Capabilities
- Vibe Coding and AI-Based Products
to
Web Designer
MeuCarroNovo
I joined at sixteen as a web designer and built the process that put an independent car dealer's site live in two hours. For three years I produced sites in volume every day: hundreds of them, each one different, sharing only the vehicle listing and viewing engine, with templates and dashboards so one dealer did not look like the next. I created that product, and I built the first three versions of the MeuCarroNovo portal itself.
Numbers
approx. 2h
to deliver a car dealer's website
approx. Hundreds
of car dealer sites, each one different
Capabilities
- Vibe Coding and AI-Based Products
05Skills
What I work with
The tools and disciplines behind the story above, grouped by what they are for.
- Data & AI Architecture
- Data Lake
- Data Warehouse
- Lakehouse
- BigQuery
- Postgres
- pgvector
- Firestore
- ELT & DataOps
- Medallion architecture
- Data lineage
- Data enrichment & bureaus
- LGPD & GDPR
- AI Engineering & Vibe Coding
- LLMs
- Context Engineering
- Prompt Engineering
- Spec Engineering
- Harness Engineering
- Multi-agent systems
- RAG
- Human-in-the-loop
- Claude Code CLI
- Gemini CLI
- Codex CLI
- Cursor
- Lovable
- Grok Build
- LangGraph
- LangChain
- Langflow
- CrewAI
- Mastra
- Vercel AI SDK
- MCP servers
- CLI tooling
- Image, Video & Audio Models
- Whisper
- DALL·E 3
- Leonardo AI
- Remotion
- FFmpeg
- Tone.js
- Generative media pipelines
- Product & Delivery
- Product management
- Product ops
- Discovery to delivery
- PLG
- Pricing & go-to-market
- Rollout models
- Design systems
- ADR documentation conventions
- Scrum
- Kanban
- GTD
- Metamodels
- SaaS & Micro-SaaS
- Automation & Low-code
- n8n
- Make
- Zapier
- Botpress
- Google Apps Script
- Google Sheets extensions
- Certifications
- Product Manager
The story, in order
25 years of building where data, software and product meet. In order, from 2001 to today.
It started with curiosity
At nine I was pixel-editing skins for Flight Simulator, Age of Empires and Quake. Taking a system apart to see how it was put together is still how I learn anything.
Sixteen, and a production line
In 2001 I joined MeuCarroNovo as a web designer, at sixteen, and built the process that put an independent car dealer online in two hours: one engine for listing and viewing vehicles, with a template and a dashboard of its own for every store. For three years I published hundreds of car dealer sites, each one different, and I wrote the first three versions of the MeuCarroNovo portal itself.
The first agency
From 2004 to 2007 I co-founded Agência da Internet with two partners, and it was the first business I ran end to end: finance, administration and operations, while being the graphic and web designer of the house. Two products came out of it. I co-founded the first Impostômetro, which the national press picked up daily, and I launched the first SóCarrão portal, today the largest automotive portal in southern Brazil.
The freelance years
From 2008 to 2010, between the agency and my return to a salaried role, I worked as a full-stack developer specialising in e-commerce, building online stores for the Curitiba market.
Planning, then product
In 2011 I managed delivery at Salsa:Mobion, then ran digital planning at IBPT and at Getz Comunicação. At IBPT I was responsible for creating and maintaining the public-interest sites Impostômetro, Lupa no Imposto, Feirão do Imposto, Empresômetro and Amiga dos Sonhos. For three years, alongside all of it, I also taught discipline, focus and habits at A Grande Escola.
The agency that became a portfolio
In 2014 I co-founded Nossa Causa, a social-transformation agency. I built a portfolio of more than 70 services, each with a scope, a deliverable and a price, with the process behind every one of them drawn in BPMN, and around them the parts an agency usually improvises: proposal, contract, sales, support and reporting. The content operation we ran on the house's own account won the Espada Ninja award from Resultados Digitais, second among performance agencies in Brazil.
Product, data and AI
Then product, all the way in: acquisition, engagement and analytics products at Trackmob, used by Greenpeace, Doctors Without Borders, the ICRC and ActionAid; Lakehouse architecture and rollout operations at RiskNow; and two products of my own, YouSendr and OnlySheets. Today I am Head of Data & AI at Liquid AI, where I lead data pipeline governance, the LGPD and GDPR compliance mapping behind Liquid AI's security and privacy framework, a company-wide AI-first culture, and the agent systems that automate critical workflows.
How I work now
Using AI for research, planning and Vibe Coding is what lets me work across infrastructure and languages I have not spent a decade inside. The method is written down as rules, contracts and gates, and what it produced can be counted.
22 R&D repositories built in 2026, each one under the same governance framework. This system was built inside it too.