Guilherme Oyakawa de Almeida
I turn messy datasets into decisions — SQL, Python and Power BI in service of the P&L. Master's in Data Science, background in Business Administration, and a habit of shipping the dashboard that actually gets used.
Numbers with a business owner attached
I'm a Brazilian-Portuguese data professional who spent the last few years turning spreadsheets into strategy — first in automotive quality on BMW supplier lines in Turin, then in independent consulting, and now full-time in data science.
After a Master's in Data Science from Rome Business School (28/30) and a Professional Certificate in Data Analytics from Imperial College Business School, I relocated to Dublin in 2026 — Portuguese citizenship in hand, ready to work across the EU without visa friction.
I like problems with a business owner attached to them: a bank losing money on bad loans, a warehouse missing its SLAs, a condo board flying blind on its budget. Give me the raw data and I'll give you the dashboard that changes the decision.
Off the clock: a lifelong Palmeiras supporter, slowly working through a backlog of sci-fi novels, and prone to disappearing down YouTube rabbit holes about geopolitics.
The stack behind the dashboards
Programming & Data
- Python (Pandas, NumPy, Scikit-learn, XGBoost)
- SQL (PostgreSQL — views, CTEs, window functions)
- SQL Server & Google BigQuery
- Scala
- ETL & data pipeline design
BI & Visualization
- Power BI (DAX, Power Query, star-schema modeling)
- Tableau
- Matplotlib / Seaborn
- Excel (Advanced / VBA)
Machine Learning & Explainability
- Predictive modeling (XGBoost, Random Forest, Ridge, k-NN)
- Clustering (K-Means, Isolation Forest)
- SHAP (explainable AI)
- Hyperparameter tuning (GridSearchCV)
Domain Expertise
- Financial risk & FP&A
- Supply chain analytics
- Customer retention & CRM
- Sports market valuation
Tools & Platforms
- Git / GitHub
- Jupyter Notebooks
- Streamlit (app deployment)
- Agile / Scrum (SFPC™ certified)
- Salesforce CRM · Cloud fundamentals (AWS, Azure, GCP)
Selected work
Five end-to-end projects — each one takes a messy dataset to a decision someone actually made.
Football Transfer Valuation Model
Built an XGBoost regression model on Transfermarkt performance data to estimate a player's fair market value, then shipped it as an interactive Streamlit app so sports directors can price transfers on data instead of hype.
Explore the full project → 61% RECALL · 89% ACCURACYPredicting Credit Default
Analyzed 30,000+ loan applications with a 21.5% baseline default rate. A k-NN classifier caught 61% of actual defaulters, and a companion Power BI dashboard flagged "G-grade" loans at a 98% default rate — turning reactive collections into proactive lending policy.
Explore the full project → POWER BI · STAR SCHEMAFP&A Dashboard for Condominium Management
Replaced a residential condominium's spreadsheet-based accounting with a Power BI financial command center — a star-schema data model and DAX-driven actual-vs-budget variance tracking the board can actually read.
Explore the full project → 5 ML MODELS · OPS DIAGNOSISDecoding an E-Commerce Delivery Crisis
Tested five ML models to predict shipment delays; none cracked 67% accuracy, because the real story was operational, not predictable. Dashboards showed "high-importance" air-freight orders were the most delayed — redirecting the fix from an algorithm to a warehouse audit.
Explore the full project → RFM · 4 EU CLUSTERSPredictive Churn Dashboard for Fashion E-Commerce
Master's capstone with Ora Fashion: while a teammate built the churn model in Python, I architected the Power BI command center — RFM segmentation, cohort analysis, and tailored retention plays mapped across four European customer clusters.
Explore the full project → EXPLAINABLE AI · IN PROGRESSGeopolitical Events & Commodity Trends Analytics
Building an explainable news-analytics pipeline with an Imperial College London collaborator — mapping which countries and companies actually drive gold, oil and metals prices, aimed at investors who can't manually track hundreds of sources.
Explore the full project →Career roadmap
Data Quality Analyst (Contract) · Coveract SRL
Inspected 2,000–3,000 BMW X5/X7 prototype door-handle components a month; identified a recurring assembly defect that cut component scrap by ~15%. Took part in 3 quality audits, including two run directly by BMW.
Career Break — Personal Goal Pursuit
Relocated to Italy and completed the Portuguese/EU citizenship process while pursuing the Master's in Data Science full-time.
Data Reporting Analyst (Contract) · Micla Engineering and Design
Coordinated durability testing for the Lamborghini Revuelto programme — tracking 20–30 tests per cycle across sensors, systems and battery performance, reporting to Lamborghini and Italdesign engineering teams.
Project Support Coordinator · Zenatur
Coordinated 56–72 Samsung deliveries a month nationwide — including the Galaxy Z Fold, Z Flip and S23 launches — maintaining a 100% on-time delivery rate throughout.
Service Delivery Coordinator · T-Systems do Brasil
Coordinated global IT service-delivery operations between Service Delivery Managers and technical teams, standardizing incident reporting across international clusters.
Store Manager · Vienna Café
Oversaw inventory and cash-flow control for a family-owned business.
Engineering Intern · Rucker do Brasil
Produced technical drawings in CATIA V5 for Peugeot, Volkswagen, Renault and 3M.
Let's talk about your data
Open to Data Analyst / Business Analyst roles across the EU — remote or Dublin-based. Reachable any time below.