AI agents
- Problem
- Repetitive, multi-step work eats your team's time.
- What we build
- Tool-using agents (LLMs, MCP) connected to your systems, with guardrails and human approval where it matters.
- Outcome
- Hours back every week, with an audit trail.
AI agency · Data engineering
AI that ships to production, built on data you can trust.
We design and build AI agents, RAG search and the data platforms underneath them, so your pilots become systems your team actually uses.
- Certified Databricks Data Engineer Professional
- Certified Google Cloud Professional Data Engineer
- Delivered at global fintech and investment platforms
Most AI projects stall for the same three reasons
Pilots that never ship
A demo works in a notebook, then dies at security review, scale or ownership.
Data nobody trusts
Agents are only as good as the data behind them. Scattered, undocumented tables produce confident wrong answers.
No governance
Who can see what? Which data feeds the model? Regulated companies need the answers before launch, not after.
What we build
RAG & knowledge search
- Problem
- Answers are buried in documents, wikis and tickets.
- What we build
- Retrieval pipelines over your own data with citations, access control and evaluation.
- Outcome
- Grounded answers your team can verify.
AI-ready data platform
- Problem
- Your data is too messy or slow to power AI.
- What we build
- Pipelines, warehouse/lakehouse, data quality and catalog on GCP, Azure or AWS.
- Outcome
- One trusted foundation for analytics and AI.
Workflow automation
- Problem
- Manual hand-offs between tools and teams.
- What we build
- LLM-assisted automations for classification, extraction and routing, integrated with what you already use.
- Outcome
- Fewer errors, faster cycle times.
Experience delivered
Selected work from the founder's career at global fintech and investment platforms.
Global payments company
Regulated data delivery and LLM/agent work
Built a Python framework reused by many data pipeline integrations, and a credit-data integration delivering critical data to 7+ external entities under regulations in several countries. Ongoing work with LLMs, RAG pipelines and MCP-based agents.
- 7+ external entities
- Multi-country regulation
Global fintech & FX
New data platform from scratch
Designed a new data platform with Airflow on Cloud Composer, dbt, BigQuery and Kafka, while keeping critical legacy services on GCP healthy. Led ClickHouse proofs of concept.
- 50+ Airflow DAGs
- GCP · Kafka · dbt
Brazilian investment platform
Data mesh, streaming and data governance
Built a data streaming platform on Azure for a data mesh architecture, on top of a Databricks lake with 2,000+ tables. Created an internal data catalog (Amundsen-based) for access requests, ownership and discovery, with IaC and audited CI/CD.
- 2,000+ tables
- 60+ active users
Google Cloud partner consultancy
Cloud data solutions for customers
Designed and implemented solutions with Pub/Sub, BigQuery, Dataflow, Dataproc and Cloud Functions as a Google Cloud Partner.
- 4+ customers
- Client-facing delivery
Engineering firm
BI platform and cloud migration
Built a BI solution on Google Cloud with ETLs from multiple sources, migrated an on-premise platform to the cloud and supported a critical platform 24/7.
- On-prem → GCP
- 24/7 critical support
How we work
01
Diagnose
A free 30-minute call to find the highest-value use case and check your data readiness.
02
Prototype
A working prototype on your real data, with clear success metrics.
03
Ship
Production deployment with security, monitoring and documentation.
04
Operate
Optional support and iteration once it is live.
Credentials
- Databricks Certified Data Engineer Professional (2025)
- Google Cloud Professional Data Engineer (2019)
- MBA, Ibmec São Paulo
- Google Cloud Partner experience
Tools we work with
- Google Cloud
- Azure
- AWS
- Databricks
- BigQuery
- Airflow
- dbt
- Kafka
- Spark
- Kubernetes
- Python
- Neo4j
- ClickHouse
- LLMs · RAG · MCP
Frequently asked questions
How much does it cost?
It depends on scope. After the free diagnostic you receive a fixed-scope proposal for a first pilot, so you know the investment before committing.
How long does a first project take?
A focused pilot typically takes a few weeks. We confirm a timeline in the proposal once we have seen your data and systems.
Is my data safe? What about LGPD and GDPR?
We design with least-privilege access, auditability and data-residency requirements from day one, and can work inside your cloud account so data stays with you.
Who owns the code?
You do. Deliverables, code and documentation are handed over to your team.
Can you work with our existing stack?
Yes. We work across Google Cloud, Azure, AWS, Databricks, BigQuery, Airflow, dbt and Kafka, among others.
Do you offer support after launch?
Yes, optionally. We can monitor, tune and extend what we build.
Let's find your first AI use case
Book a free 30-minute diagnostic, or tell us what you want to automate and we will reply within one business day.