10+ years in analytics, the last seven building end-to-end executive dashboards — from data modeling and governance through to delivery in Tableau, Power BI, Looker and Looker Studio — for Marketing, Commercial, Finance and Planning teams. 7+ strategic dashboards for enterprise clients including Serasa Experian, Johnson & Johnson, DIA and Alper. Currently Senior Data Analyst and Team Leader at Keyrus, leading 12 analysts.

My route into data was not a bootcamp. I was a clinical nutritionist coordinating IT in hospital nutrition, writing SQL and building analytical views for published clinical research — which is where the interest started, and why I moved into data deliberately in 2019. It left me with two habits that still shape how I work: validate the source before trusting the number, and explain the result to someone who does not share your vocabulary.

BI & DataViz

DAX · Power Query (M) · Power BI Service · performance optimization · LookML

Data & Cloud

subqueries · CTEs · window functions · data modeling · Data Lake · data governance

Analytics & Business Tools

Metrics I have owned in production

LFL · YoY · MoM · revenue projections · market share · promo share · gross margin · PDV margin · average ticket · GSUB · churn · ROI · CAC · LTV · conversion rate · CTR · CPC · CPA

• Executive dashboards — end-to-end ownership: requirements → data modeling → governance → delivery in Power BI, Tableau, Looker or Looker Studio, including performance optimization and the report lifecycle through to production
• Revenue & customer analytics — RFM segmentation, cohort retention, LTV, churn, discount and margin diagnosis
• Marketing & growth analytics — acquisition funnel, channel performance, campaign KPIs (CAC, LTV, ROI, CPA, CTR, CPC)
• Commercial & retail analytics — like-for-like, market and promo share, mix shift, ABC assortment, average ticket
• Team leadership — 12 analysts across enterprise accounts: mentoring, review standards and people development (playbook)
• Technical documentation — requirements, business rules and data dictionaries, treated as part of the deliverable

Five published dashboards on Superstore, built around one argument: we are growing, and growth is not profit

• Filled profit maps (diverging palette centred on zero) surface loss-making states: 10 of 49 lose money
• The money is sized, not only diagnosed — US$323K given up to discounting against US$156K lost on unprofitable orders
• Every account ranked with its margin, because a segment is not something anyone can act on
• Table calcs: Pareto cumulative share, YoY growth, moving averages; click-to-filter across every page
• Stack: Tableau (maps · table calcs · dashboard actions) · packaged .twbx in the repo

A six-page Power BI report on 4 years of retail data — growth, customer value, product lines, and a discount what-if

• Star schema shaped in Power Query (M), 44 DAX measures, page navigation and a what-if parameter
• Discount impact — orders above 30% off have an 83% loss rate; break-even sits near 20%
• Customer acquisition told as a trend rather than a number: 595 new customers in 2015 down to 11 in 2018
• Time intelligence written as explicit date arithmetic, and the repo documents why: the functions that need a marked date table fail silently when the calendar overshoots the data
• Stack: BigQuery SQL (8 queries) · Power BI (DAX · Power Query) · shipped as a .pbip project

The metrics a commercial director actually asks for — on 5.2M real invoice lines across 2,110 stores and US$875M of wholesale revenue

• Like-for-Like with a bridge that reconciles: headline +4.12% is only +3.01% LFL — over a quarter of the growth was store count, and 97 lost accounts appear nowhere in the headline
• Market share within category — the competitive set, not total revenue — with movement in share points: Straight Bourbon's leader lost 3.19 pp while the category grew 7.6%
• Promo share detected from price deviation (no promo flag exists): share fell by a third while discount investment rose 38.5% — narrow deep discounting replaced broad shallow, costing 3.65 pp of margin
• Category growth bridge — Single Barrel Bourbon grew +70% and contributed 0.22 pp; American Vodkas grew +7.7% and contributed 1.12 pp. Rate is not impact
• ABC curve — 28.1% of stores carry 80% of revenue, at a 632-item assortment against 111 in the C band
• Stack: BigQuery SQL (6 queries · window functions · growth bridges) · live public dataset

End-to-end analysis of 94K+ orders from theLook, a global e-commerce marketplace (live BigQuery public dataset)

• Revenue growth from US$73K (2019) to US$2.1M (2025), +54% YoY, with MoM and 3-month rolling average
• RFM segmentation (per user_id) — Champions (8.6% of customers) drive 21.3% of revenue; top segments = 55%
• Cohort retention matrix (heatmap) — 100% → ~5% at M1, stabilizing ~2% with a durable long tail
• Geographic & category analysis with profit margin (52% blended) — China/US/Brazil ≈ 70% of revenue
• Stack: BigQuery SQL (9 queries · window functions) · Looker Studio · public dataset (fully reproducible)

Google Analytics 4 analysis on the raw BigQuery export — acquisition, engagement, conversion funnel, and revenue (270K users)

• Handles GA4's nested schema: sessions from event_params, channel grouping from traffic_source, products via UNNEST(items)
• Full purchase funnel (view_item → add_to_cart → begin_checkout → purchase) — biggest drop-off is view→cart (only 20.5% add)
• Channel quality vs volume: Organic leads on users, Referral converts best (1.85%); Mobile drives 40% of revenue; US = 44%
• Stack: BigQuery SQL (8 queries · nested/UNNEST · window functions) · Looker Studio · public GA4 export

A working Excel workbook on 57,542 order lines — the tool business stakeholders actually read numbers in, treated as a first-class analytical surface

• Distinct-order counting in O(n) — the textbook SUMPRODUCT(1/COUNTIF()) is O(n²) and hangs on 57K rows; an adjacency flag on a pre-sorted key does it in one pass
• Revenue-weighted margin via SUMPRODUCT, top-N Pareto via INDEX/MATCH/LARGE, two-way interactive lookup with dropdowns, PERCENTILE.INC distribution
• Power Query (M) loader with locale-proof en-US typing — a pt-BR machine otherwise reads 79.95 as 7995
• Growth diagnosis: +54.8% revenue came from order volume (+52.6%), not basket size (+1.5%), at flat margin
• Stack: Excel 2016-compatible formulas · Power Query (M) · BigQuery SQL

How I run a data/BI team — written from leading 12 DataViz and Business Analysis analysts across enterprise accounts

• Dashboard brief that opens with the question ending most projects early: what decision will this change, and who makes it? Forces revenue recognition, FX, fiscal calendar and YoY basis to be settled before the build, not in a meeting where two numbers disagree
• Review checklist ordered by damage, not ease — including the most common silent error in BI: blank and zero rendering identically, so a month with no sales looks like a month with no data
• 30/60/90 onboarding built around time-to-unsupervised-delivery, front-loading business context over tooling
• Intake and prioritisation that makes the capacity trade-off visible instead of refusing work — plus the part everyone skips, dashboard retirement
• Stack: requirements templates · review standards · people development · Trello/Asana workflow

Keyrus · Apr 2023 – Jun 2026 · Rio de Janeiro

• Team Leader (Feb 2025 – Jun 2026) — led 12 DataViz and Business Analysis analysts across multiple client accounts: quality of delivery, business alignment, mentoring and performance management.
• Senior DataViz Consultant (Apr 2023 – Jun 2026) — strategic multi-page dashboards for Serasa Experian, Johnson & Johnson, Alper and DIA across Marketing, Commercial, Finance and Planning. Full ownership: requirements, modeling, SQL in BigQuery, Data Lake integration, governance, and delivery in Tableau, Power BI, Looker, Looker Studio and Databricks. Analyses that supported supplier renegotiations, product mix reviews and campaign investment reallocation.

RecargaPay · Dec 2019 – Nov 2022 · Rio de Janeiro

• Marketing & Growth Analyst (Jul 2021 – Nov 2022, promoted from Commercial Analyst) — B2B and B2C behaviour and sales performance for a fintech SaaS product, using SQL, Qlik View and Looker Studio. Funnel prioritisation across activation, conversion and retention, with weekly executive readouts.

Clinical nutrition & research · 2009 – 2019

• Coordinated IT at Nutricritical and Nutriclínica, including the rollout of an in-house nutrition system for hospitalised patients. Analysed clinical data with SQL and SPSS for studies published in Brazil and abroad. Collaborating lecturer, Stomatherapy specialisation, UERJ (2018–2019).

MBA in Marketing — FGV (Fundação Getúlio Vargas) · 2017–2018 BSc in Nutrition — UNESA · 2004–2009

SQL Fundamentals — DataCamp · Learn SQL — Codecademy · SQL for Data Analysis — Udacity · Tableau Certified Data Analyst Training — Udemy · Advanced Excel and Power BI — Udemy · The Data Analytics Course — Udemy · Complete Machine Learning & Data Science — Udemy · Introduction to Data Science — Data Science Academy · Project Management — FGV Online · Statistical Analysis with SPSS — Udemy

Portuguese native · English professional working proficiency · Spanish limited working proficiency

Every project here runs on public or synthetic data and is reproducible end to end: the SQL, the data model, the measure definitions and the build documentation are all in the repo. Client work is confidential and appears nowhere in this portfolio.