Data Analyst · Analytics Engineer

André Matiello

I turn raw data into models a business can trust and act on.

A modern data portfolio built on real market data: SQL, dbt, Python and BI, with a bit of AI used with judgment, not as a crutch.

Projects

Built on real market data

Data Analyst at the core, with range across Analytics & Data Engineering. Click any card for the full write-up: context, architecture, challenges and results.

AE · 01Live

Declarative ingestion with dlt, layered dbt transformation and a dimensional model, built on DuckDB in dev and portable to Snowflake.

Modern Data Stack ELT

✓ 7 models · 22 tests · 0 errors · 5,010 rows · green on DuckDB + Snowflake
dltdbtDuckDBSnowflakeSQL
DE · 06Live

Medallion layers and a Kimball star schema in pure SQL, with PL/pgSQL ETL over dirty CRM + ERP extracts, reproducible with one command.

SQL Data Warehouse on PostgreSQL

✓ 18 quality checks green · 60,398-row fact · revenue reconciled across layers
PostgreSQLPL/pgSQLSQLDocker
DA · 02In progress

Multi-page BI dashboard with cross-filters and a 2-minute narrated demo on real market data.

Interactive Dashboard + Narrative

TableauPower BIStorytelling
DA · BA-01Live

Segmentation and revenue-at-risk quantified with SQL window functions behind a live, cross-filterable dashboard.

Customer Churn: Revenue at Risk

✓ 9 quality checks · 26.54% churn · $58,277/mo top segment risk
SQLPostgreSQLWindow functions
DA · BA-02In progress

Excel as a real analytics engine: Power Query ETL, a Power Pivot star schema, DAX measures and what-if scenarios.

Excel Analytics

Power QueryPower PivotDAX
DA · BA-04Live

A real quasi-experiment on 95,824 Olist orders: hypothesis test, confidence interval, power analysis and a distance-confound check behind a clear call.

Delivery SLA Impact on Customer Satisfaction

✓ 48.2pp satisfaction gap · p<0.001 · power=1.0 · holds across every region & distance quartile
Pythonscipystatsmodels
DE · 05 ⭐Live

A FastAPI proxy over LLM APIs feeding a medallion lakehouse that answers: which team is burning the AI budget?

LLM Token Gateway + AI FinOps

✓ 42/42 tests green · 39/39 dbt (idempotent, 12,934 rows) · anomaly injected & caught (z=565)
FastAPIdbtDuckDBFinOps
AE · 05In progress

dbt docs on an S3 static site, and a Streamlit dashboard on EC2 reading AE-01's market mart through a least-privilege IAM role, with Terraform end to end.

AWS Cloud Foundations

TerraformAWS S3EC2IAM
DE · 07In progress

A governed medallion lakehouse for ICU clinical analytics (MIMIC-IV + eICU), with a canonical clinical model layer that harmonizes two hospital data sources before anything reaches a data product.

US Critical Care Lakehouse

DatabricksDelta LakeUnity CatalogPySpark
TRANS · 04Live

The One Billion Row Challenge in Python: DuckDB, PyArrow and Polars on a real 1-billion-row aggregation, including a reproduced out-of-memory failure.

Big Data Engines Benchmark

✓ DuckDB 22.73s · PyArrow 1,185s · Polars OOM-killed at 12.51 GiB (kernel log)
DuckDBPyArrowPolarsPython
DE · 08Live

A modular, stage-tested Python ETL pipeline: Excel workbooks in, governed Parquet and Excel out, with a CLI runner that tests and logs every stage.

Modular ETL Pipeline Engineering

✓ 4/4 tests green · 50 workbooks → 500 rows · verified from a clean clone
PythonpandasPyArrowpytest
Skills

The skill set behind the projects

A working subset of 25 skills pulled from a longer list, kept because each one shows up in a project above, not because it looks good on a resume.

Query & Programming

SQLPythonETL

CTEs, window functions and API-to-warehouse scripting: AE-01, BA-01, DE-06, DA-01.

Modern Data Stack

Data PipelineData Warehousedbt (Data Build Tool)Data LakeBig DataData Quality

Layered dbt models over a warehouse, plus the lakehouse and orchestration side: AE-01, DE-01, DE-02, DE-05, DE-06. Big Data engine trade-offs at scale: TRANS-04.

Cloud

Amazon Web Services (AWS)

S3, EC2, Lambda and least-privilege IAM behind the orchestrated and infra-as-code pipelines: AE-05, DE-01, DE-04, DE-05.

BI & Visualization

Microsoft Power BIMicrosoft Power QueryMicrosoft ExcelDashboardsData VisualizationStorytelling

From a Power Pivot star schema in Excel to a cross-filterable web dashboard: BA-01, BA-02, DA-02, DA-03.

Statistics & Experimentation

StatisticsStatistical InferenceProbability Theory

Hypothesis testing, confidence intervals and power behind every A/B call: BA-04.

Business Analysis

Data AnalysisBusiness IntelligenceBusiness RequirementsRequirements GatheringData CleaningReport Building

Translating a business question into requirements, clean data and a recurring report: BA-01, BA-02, BA-03, DA-04.

Blog · Writing

Notes from building

The decisions behind the projects, and what the data says about this market.