╔══════════════════════════════════════════════════════════════╗
 ║   YULIA CARVALHO  ·  DATA ANALYST & DATA ENGINEER            ║
 ║   Lisbon, Portugal  ·  Available · Open to Remote / On-site  ║
 ╚══════════════════════════════════════════════════════════════╝

I spent fifteen years in high-impact research environments - molecular biology, biomedical science, large-scale diagnostics - where the underlying work was always the same: design a rigorous process, gather imperfect data, extract reliable signal from noise. That instinct transferred cleanly into data engineering and analytics.

What I enjoy most is what I call forensic EDA: a structured, investigative approach to exploratory analysis. Analysis, for me, starts before dashboards or models — with questioning the integrity of the data itself. Cleaning inconsistencies, validating assumptions, identifying anomalies, tracing duplicates, resolving cross-field conflicts. I approach datasets the way I once approached laboratory experiments: carefully, skeptically, and methodically, until the underlying logic becomes clear.

I'm drawn to the engineering side of analytics: building reproducible pipelines, defining clear data contracts, designing layered transformation architectures, and documenting logic so that analytical outputs can actually be trusted downstream.

15+ years research · 12 peer-reviewed publications · Lisbon, PT
EN · PT · RU

Python BigQuery dbt Core GitHub Actions Looker Studio

End-to-end data engineering pipeline ingesting hourly weather observations from 222 meteorological stations across Portugal via the IPMA public API. Implements a layered architecture (raw → staging → mart) with automated data quality testing, incremental batch loading, and daily scheduling.

• Append-only raw layer with ingested_at timestamps for full auditability and root-cause tracing
• Dual-table ingestion pattern reflecting different update semantics (append vs. truncate-reload)
• Two-layer dbt transformation: type casting, sentinel value nullification, wind direction decoding, deduplication on natural key
• 8 automated dbt tests covering null constraints, timestamp validity, and measurement field ranges — pipeline fails automatically on violation
• Full pipeline orchestrated via GitHub Actions cron with failure notifications and run logs

~4 151 observations/day · 222 stations · 8 quality tests

→ Repository

SQL BigQuery R Kaggle

Capstone project for the Google Data Analytics Certificate. A comprehensive operational study of Brazil's Olist marketplace diagnosing the structural drivers of customer dissatisfaction, logistics failure, and seller performance risk.

• End-to-end pipeline connecting BigQuery to a statistical environment, integrating transaction, logistics, and sentiment data
• Multiple linear regression isolating delivery delay as the dominant predictor of customer satisfaction
• Composite seller risk scoring system weighing historical lateness, delay severity, and business value at risk
• Origin–destination route analysis distinguishing individual seller inefficiencies from systemic regional constraints
• Interaction regression measuring how seasonal shopping events amplify the impact of delivery delays

100K+ orders · 18 business questions · 1 operational dashboard

→ Notebook · → Repository · → Dashboard

Python SQL Google Colab Looker Studio

Graduation project for the Le Wagon Data Analytics Bootcamp. Structured environmental analysis examining the relationship between land-use patterns, protected-area coverage, and the Biodiversity Intactness Index across 27 EU countries.

• Multi-source integration: land-use, protected-area, and biodiversity datasets into a unified analytical framework
• Cross-country structural pattern comparison identifying recurring drivers of biodiversity outcomes
• Looker Studio dashboards translating findings into accessible, decision-oriented visual communication

27 EU countries · 5 data sources · team graduation project

→ Dashboard / Report

Languages SQL — CTEs, window functions, validation logic · Python — data analysis, pipeline scripting · R — statistical modeling, reporting · HTML5 / CSS3

Data Engineering BigQuery · dbt Core (raw → staging → mart) · GitHub Actions · Airflow (concepts, DAGs) · Fivetran (ELT concepts) · DuckDB

Analysis Libraries pandas · NumPy · scikit-learn · Prophet · tidyverse / lubridate · BeautifulSoup

Visualization & BI Looker Studio · Power BI (DAX, data modeling) · matplotlib / seaborn · Plotly · ggplot2

Workflow Git / GitHub · Jupyter / Colab · VS Code / RStudio · Kaggle · Bash / CLI

• Google Data Analytics Professional Certificate · Google / Coursera · Jan 2026
• Data Analytics Bootcamp (DGERT-certified) · Le Wagon · Nov 2025
• PRINCE2 Foundation · PeopleCert / AXELOS · Jan 2024
• ETL in Python and SQL · LinkedIn Learning · Feb 2026

12 peer-reviewed papers across cardiovascular biology, molecular biology, and bioinformatics.

• Cardiovascular Research (2024) · Developmental Cell (2022) · PNAS (2021)
• Circulation (2017) · Circulation Research (2014, 2018) · Nucleic Acids Research (2012)

→ Full list on ORCID · → Full list on Portfolio