Power BI · SQL · Python · Tableau · Excel | Vancouver, BC, Canada

I am an analytical and results-driven professional with over 6 years of experience in data management, payroll analysis, project coordination, data validation, and research and business analysis across education, public administration, and research sectors.

I am a PMP-certified professional with a Master's in Analytics, specializing in transforming complex data and processes into clear, concise reports, visualizations, and deliverables that inform decision-making and enhance operational efficiency. I find the errors, anomalies and cost leaks hidden in large records, then turn what I find into SQL checks, dashboards and models that others can rerun. I am adept at coordinating multi-stakeholder communications and supporting organizational change initiatives.

• 🔭 Current Focus: Research analysis at Participedia, and a four-project public-data analytics program run as one PMP-managed portfolio (see the roadmap below).
• 💡 Goal: To leverage data science and project management to drive organizational efficiency and strategy.
• 📫 How to reach me:

Four public-data analytics projects planned and delivered as one program, from data acquisition to published dashboards. Every task is tracked with its phase, real start and target dates, and a closed issue in its repository.

Project	What it shows	Tools	Links
🗳️ Participation Design Advisor	Which designs of citizens' assemblies, participatory budgets and consultations go with documented impact, from 2,260 Participedia cases	Python, logistic regression, JavaScript
🦺 Site Safety Early Warning	Which zone of a construction site is most likely to hurt someone in the next five work days, and the controls to apply before the shift (BC, Ontario, Alberta)	Python, gradient boosting, JavaScript
📮 Canada Post turnaround analytics	Cost-saving and mailbox-siting model across 1,646 forward sortation areas	Python, Census 2021, optimization
📑 Change Orders: federal contract cost growth	351,872 contracts, award-day risk model, PMP project controls (EVM, PERT, Monte Carlo)	Python, scikit-learn, open data
🚌 Next Shift: Ontario autoworker job match	Interactive tool ranking jobs for laid-off autoworkers from 89,557 Job Bank postings	Python, JavaScript, open data
🌍 AI & Big Data for African Tax Reform	Ghana field survey of 53 informal workers and 21 professionals; readiness segments; paper check	Python, SciPy, survey design
✈️ U.S. flight delay pipeline	End-to-end pipeline on 4.1M flights; reproduces all 96 published DOT figures	Python, SQL, SQLite
⚡ Canadian household energy, 2011–2021	Power BI dashboard cross-checked against a SQL analysis	Power BI, DAX, SQL
📊 Executive Portfolio Diagnostic Suite	PMO dashboard on 1M+ synthetic records: cost variance, earned value, burn-down	Tableau

More detail on each project is on my portfolio site.

Ranks the design choices behind citizens' assemblies, participatory budgets and public consultations by how often each one appears in cases with documented impact, across 2,260 Participedia cases. A logistic regression on case features drives an interactive tool that returns a recommended design for a given purpose, scale and budget, with the evidence and the uncertainty behind each recommendation. A companion Power BI report relates country governance indicators to the same case base. Independent analysis of open Participedia data.

Ranks which zone of a construction site is most likely to injure someone in the next five work days, from signals a site already records: near misses, hazard observations, PPE and housekeeping audits, new workers, overtime, trade congestion and the weather forecast. Across 39,990 held-out zone-days on 30 sites in three provinces, the top-risk band covered 9% of zone-days, carried 3.5 times the event rate and captured 31% of all incidents (AUC 0.754 against 0.662 for a crew-size baseline). Each flagged hazard is paired with the control and the governing provincial provision. The same approach was then tested on 948,000 real New York City site-weeks from open records. Daily site logs are simulated, because contractor logs are not public; they are calibrated to published WorkSafeBC, AWCBC, Statistics Canada and Alberta OHS statistics and run on real Environment Canada weather. Not affiliated with any regulator or employer.

Combines Canada Post delivery-mode data with the 2021 Census and the National Address Register (10.7M buildings) across all 1,646 forward sortation areas. It estimates the saving from converting door-to-door delivery to community mailboxes (about $487M a year against $719M of mailbox capital, a model estimate) and ranks the areas to convert first. It also sites mailboxes on real street networks in three case-study areas, and a parcel-churn model flags customers at risk of leaving. Independent analysis, not produced by or affiliated with Canada Post.

Follows 351,872 Government of Canada contracts ($10K+, awarded 2017–2023) through three years of amendments: their value grew from $83.7B to $98.7B (+18%). An award-day model (AUC 0.81) flags contracts likely to grow 25% or more using only what is known at signing. A project-management layer treats each contract as a project: earned value at close (CPI 0.85), reference-class forecasting, PERT estimates, a Monte Carlo contingency backtested out of time, and an expected-monetary-value risk register.

For workers laid off at Stellantis Brampton and GM CAMI Ingersoll. A worker picks a plant job and the longest drive they'd make. The tool then ranks Ontario occupations by skill fit (ESDC's OaSIS ratings), Job Bank openings within a rush-hour drive, pay, and training time, and lists nearby college programs and funding. A planning view estimates how many laid-off workers local openings could absorb. Independent analysis from public data. Job postings are not hires, and the workforce mix and the share of openings workers can win are adjustable assumptions.

The evidence base for my GCA paper. 53 hand-coded informal-sector questionnaires, a survey of 21 tax and finance professionals and MTN Ghana's answers show that 98% of workers use mobile money and 85% say they should pay tax, yet only 36.5% have heard of AI. Awareness, not income, decides acceptance. Workers split into three readiness segments (26% ready, 35% persuadable, 39% resistant), and 27 of the paper's 29 figures are confirmed against the coded forms.

Loads seven monthly Bureau of Transportation Statistics files (4,120,342 domestic flights, 13 carriers, 357 airports) into SQLite and runs 12 data-quality checks. It then recomputes on-time rates for every carrier and month. All 96 recomputed figures match the Department of Transportation's published Air Travel Consumer Report within its rounding band. A Simpson's paradox check of all 78 carrier pairs found rankings that reverse when carriers are compared at the same airports.

A Power BI dashboard on Statistics Canada data for 10 provinces, built around a 70.84 GJ national benchmark. A matching SQL analysis of 989 provincial observations ranks the provinces (Alberta is highest at +11.4 GJ) and reproduces the dashboard figures, so the two tools cross-check each other.

A Tableau PMO diagnostic on 1M+ synthetic records across 50 projects. It shows actual cost against earned and planned value, portfolio cost variance, and resource use.

A Macro-micro–Hybrid Ensemble Framework for Institutional Governance: a hybrid CatBoost and XGBoost ensemble that classifies 2.26K+ Participedia cases at 76% accuracy, against baseline models near 52%, and forecasts annual case volumes to 2030. The paper grew out of the Participation Design Advisor project and shares its analysis and data. Submitted to Taylor & Francis (September 2026); under review.

Leveraging Artificial Intelligence and Big Data for African Tax Reform: AI adoption and revenue potential in Africa's informal sector, combining predictive models with a consented field study of financial professionals and informal-sector workers. Accepted by the Global Conference Alliance Inc. (March 2025); presentation pending.