10+ Years Operations & Analytics | Power BI · SQL · Python | Ex-Team Supervisor → Data Analyst

📍 Dublin, Ireland | 🇧🇷 Brazilian | 🔓 Open to Work

I'm not a fresh graduate learning to code.
I'm a seasoned operations professional who's been living in data for over a decade.

I've supervised collections teams, managed project analytics, presented to C-level executives, and built Power BI dashboards that drove real business decisions — not just portfolio pieces.

What makes me different:

• ✅ 10+ years operational experience across collections, sales, project management
• ✅ Led teams that lived and died by the KPIs I tracked
• ✅ Built production dashboards used by real businesses (not just classroom projects)
• ✅ Improved process efficiency by 70% through structured data analysis
• ✅ Published scientific researcher in Environmental Sciences (Unifesp)
• ✅ Presented insights to C-level — I know how to translate data into decisions

Now I'm formalizing what I've always done intuitively — turning messy operational data into clear, actionable insights.

Technology	Experience	Real-World Applications
SQL	1+ years 500+ lines in repos	• BigQuery schemas & star models • Fraud detection queries (P99 <1s) • Complex JOINs & window functions • Performance optimization
Python	1+ years 400+ lines in repos	• ETL pipelines (Apache Beam) • Data validation & quality checks • PII tokenization (GDPR compliant) • Pandas/NumPy data analysis
Power BI	4+ years Production dashboards	• Built dashboards from scratch • Executive reporting for C-level • 70% process efficiency improvement • DAX, Power Query, data modeling
Excel	10+ years Advanced level	• Complex formulas & pivot tables • VBA macros for automation • Financial modeling & forecasting • Operational reporting templates
GCP	Academic + Projects Production patterns	• BigQuery lakehouse architecture • Pub/Sub, Dataflow, Cloud Storage • Vertex AI ML pipelines • Multi-zone medallion design

My professional experience extends far beyond public repositories:

• SQL: 1+ years writing complex queries, stored procedures, database optimization
• Power BI: 4+ years building production dashboards for executive decision-making
• Python: 1+ years for data automation, ETL processes, and analysis (Pandas, NumPy)
• Excel: 10+ years with advanced formulas, pivot tables, and VBA macros
• Team Leadership: Supervised operations teams, presented to C-level stakeholders

GitHub repos show recent projects. Enterprise work was in private repositories.

May 2023 - Present

• Designed operational reporting templates in Excel, standardizing performance tracking
• Analyzed guest flow & occupancy patterns, identifying inefficiencies
• Translated raw operational data into actionable insights for management

July 2022 - April 2023

• Built Power BI dashboards from scratch for executive reporting
• Presented insights directly to leadership, bridging data and strategy
• Drove 70% improvement in post-implementation efficiency via data analysis

February 2018 - June 2022

• Monitored operational & financial KPIs across collections and sales teams
• Developed management reports improving decision-making efficiency
• Led customer segmentation analysis to enhance targeting strategies

2012 - 2017

• Published scientific research project analyzing environmental data
• Applied statistical methods and geospatial analysis (ArcGIS, AutoCAD)
• Foundation in evidence-based decision making

Production-Grade Medallion Lakehouse for Cross-Border Fintech

Designed a multi-zone data platform for a fictional Ireland-Brazil fintech processing 55,000+ transactions/month:

• ⚡ 99.9% SLA compliance with P99 <1s fraud detection latency
• 🔒 GDPR & PCI-DSS compliant with column-level encryption
• 💰 60% TCO reduction vs legacy ERP systems
• 🏗️ Five-zone medallion architecture: Pub/Sub → GCS → Dataflow → BigQuery → Looker

Tech Stack: Google Cloud Platform, BigQuery, Apache Kafka, Dataflow, Vertex AI, dbt, Great Expectations

Key Innovations:

• Hybrid ingestion (streaming + batch + micro-batch)
• Kimball star schema with SCD Type 2 customer modeling
• Real-time fraud detection via BigQuery ML (AUC >0.85)
• Federated governance with DataHub lineage tracking

Code Highlights:

• 📄 500+ lines SQL: Star schema design, fraud detection queries
• 🐍 400+ lines Python: ETL pipelines, data validation, PII tokenization
• 📊 Complete documentation: Technical specs, architecture diagrams

📄 Read Full Technical Documentation

An End-to-End Power BI Investigation That Builds an Argument, Not Just Charts

A narrative-driven diagnosis of why a call center closed 2020 with customers in distress — analyzing 13,055 interactions across 4 channels, 6 contact reasons, and the full United States:

• 🚨 NPS −69 with 51% negative sentiment surfaced as a systemic, not local, problem
• 🔍 Five-hypothesis investigation (channel, reason, geography, seasonality) — each ruled out with evidence
• 🧮 Null-safe DAX NPS measure — corrected a −78 → −69 distortion caused by blank surveys counted as detractors
• 🎯 Prioritized recommendations: fix efficiency (61% of calls >20 min) + deflect 77% money-related demand

Tech Stack: Power BI, DAX, Power Query, dimensional (star schema) modeling

Key Highlights:

• Drill-through detail pages by U.S. state and custom sidebar navigation
• Honest visualization: zero-anchored axes and CSAT shown on its true 0–7 scale
• Data storytelling structured as problem → eliminated suspects → root cause → action

Cover	Overview ("The Alarm")
Root Cause	Recommendations

📄 Explore the Full Report

Evidence-Based Visual Analytics on 8,494 Retail Products — Graded 88/100

A visual analytics study testing whether the beauty industry's price-equals-quality promise holds up, across a real Sephora catalogue of 8,494 products from 304 brands:

• 🎯 No price–quality link: a $15 moisturiser can out-rate a $200 cream — value beats cost
• 🧮 Custom metrics built from scratch: Value Score (rating per dollar) and Review Density (reviews per love)
• 📊 10 business questions → 10 purpose-built charts, each chart type matched to the data
• 🎨 Grounded in visualization theory: Cleveland & McGill (length over angle), Tufte (data-ink ratio), colourblind-safe palette

Tech Stack: Microsoft Excel, VLOOKUP, IF logic, PivotTables, star-schema modeling, Kaggle dataset

Key Highlights:

• Star-schema workbook — fact table plus a price-range dimension over 8,494 products with 8 calculated fields
• Budget products dominate rating-per-dollar; limited editions win loves but not ratings
• Viral products show broad but shallow engagement — many loves, few written reviews

Top Brands by Loves	Average Rating by Price Bucket
Category Distribution	Best Value (Rating per Dollar)

📄 Explore the Full Project

A BigQuery Analytics Consulting Study, From Star Schema to Revenue Strategy — Graded 88/100

Acting as a Senior Data Consultant on a retail client's BigQuery warehouse, turning six SQL-driven insights into five costed revenue strategies:

• 🏛️ Four-layer medallion architecture (raw → clean → warehouse → data mart) feeding a governed vw_sales_summary single source of truth
• ⭐ Retail star schema: fact_sales with seven dimensions, built for OLAP aggregation
• 🧮 RFM segmentation in BigQuery SQL using NTILE(5) scoring and CASE logic to split customers into Champions, At Risk and Lost
• 💶 Insight-to-revenue roadmap: five strategies with projected uplift (5–20%) and effort ratings

Tech Stack: Google BigQuery, SQL (window functions, CTEs, NTILE), Star Schema, Medallion Architecture, Looker Studio

Key Highlights:

• Six analytical insights powered by RANK(), NTILE(5), DATE_DIFF and CTE-based queries
• Revenue-concentration analysis (top 1% of customers ≈ 28% of revenue) framed as a retention opportunity
• Future-state design: BigQuery ML (churn, ARIMA_PLUS forecasting), streaming and a semantic layer

Warehouse Architecture	Retail Star Schema
Revenue Growth Strategy Map	Executive Dashboard Mockup

📄 Explore the Full Project

Applied Machine Learning — Graded 78/100

A dual binary classification project predicting vaccine uptake from the 2009 U.S. National H1N1 Flu Survey (26,707 respondents, 35 features) to help public health teams target outreach:

• 🎯 ROC-AUC up to 0.86 (Random Forest, H1N1) following a full CRISP-DM pipeline
• ⚙️ GridSearchCV tuning across C, penalty, and solver with 5-fold stratified cross-validation
• 💶 Business-impact framing: false positive < €10 vs false negative > €1,000 → recommends lowering the threshold to 0.3–0.4 to maximize recall
• 📊 10 EDA visualizations; doctor recommendation identified as a top predictor for both targets

Tech Stack: Python, scikit-learn, pandas, NumPy, matplotlib, seaborn, missingno

Key Highlights:

• Median/mode imputation, one-hot encoding, StandardScaler, stratified 60/20/20 split
• Logistic Regression vs Random Forest comparison across both targets
• Conceptual deployment design: REST API + EHR integration with weekly retraining for concept drift

📄 Read the Full Analysis

Business Intelligence & Data Warehousing — Graded 85/100

Design and critical evaluation of an end-to-end pipeline that moves a medium-sized retail organization from a strained transactional MySQL system toward a governed analytical ecosystem:

• 🏛️ Six-layer architecture mapped onto the Medallion pattern (Bronze → Silver → Gold)
• ⭐ Kimball star schema with Type 2 Slowly Changing Dimensions for historical tracking
• 🛡️ Layered data quality framework: duplicate detection, null validation, referential integrity, domain checks
• ⚖️ Critical evaluation of real trade-offs: batch vs streaming, query performance vs schema flexibility, cost vs performance

Tech Stack / Concepts: Data Warehousing (Kimball & Inmon), Medallion Architecture, Star Schema, OLTP vs OLAP, SQL, MySQL

Key Highlights:

• Raw vs clean staging layers enabling validation checkpoints and pipeline restart
• Metadata and data lineage treated as first-class governance concerns
• "Near benchmark quality" — one of the strongest submissions per lecturer feedback

📄 Read the Full Report

City Colleges Dublin | 2025 - 2026

Unifesp - Universidade Federal de São Paulo | 2012 - 2017
Published scientific research project

• ✅ Microsoft Power BI for Business Intelligence and Data Science
• ✅ General Data Protection Law (GDPR/LGPD)
• ✅ Lean Six Sigma White Belt
• ✅ Integrating AI with Other Skills

• 🎯 Business-first mindset: I don't just query data — I solve business problems
• 👥 Team leadership: Supervised teams, coached performance, drove results
• 📊 C-level communication: Presented insights to executives who don't speak SQL
• 🔄 Process improvement: 70% efficiency gains through data-driven redesign
• 🌍 Cross-cultural: Worked across Brazil, Ireland, hospitality, finance, environmental sectors
• 📈 Real production experience: Dashboards that ran businesses, not just portfolios

1. I've managed the teams that use the reports I build
2. I've defended the numbers in boardroom presentations
3. I know what "good data" looks like because I've seen what bad data costs
4. I understand operations — data isn't abstract to me; it's how businesses run

I'm seeking Data Analyst, Business Analyst, or BI Analyst roles where I can:

• ✅ Turn operational chaos into clean, actionable insights
• ✅ Build dashboards that actually drive decisions (not just look pretty)
• ✅ Bridge the gap between technical teams and business stakeholders
• ✅ Apply 10+ years of operational intuition to data problems

Location: Dublin, Ireland (or remote within EU)
Work Authorization: Stamp 1G (full work rights in Ireland)
Languages: Portuguese (native), English (fluent), Spanish (working proficiency)

📧 Email: marcelo.dafonsecaoliveira@gmail.com
📱 Phone: +353 83 834 2102
🔗 Portfolio: github.com/mfonsecaoliveira

2008-2010  │ Administrative Assistant → Air Ticketing Analyst (GOL Airlines)
2013-2016  │ President, Junior Enterprise (Environmental Sciences)
2016-2017  │ Environmental Data Analyst (Waterloo Brasil, Geojá)
2017-2018  │ Account Receivable Analyst (Full Time Soluções)
2018-2022  │ Credit Control & Commercial Supervisor
2022-2023  │ Project Analyst (70% efficiency improvement)
2023-Now   │ Operations Coordinator (Dublin) + HND Data Analytics
2026       │ → Data Analyst (YOU?)

"Most data analysts can build a dashboard. Fewer know what it feels like to stand in front of a management team and defend the numbers — or to lead a team that lives and dies by the KPIs on that screen.

I spent over a decade doing exactly that. Now I'm formalizing what I've always done intuitively: turning data into decisions that matter."