Six Engagements, One Discipline: Evidence Over Assertion

Classified 7,039 Job Listings by Skill Evidence, Surfacing 1,134 Hidden Data Roles

DataCareer App sourced Australian data-role listings by matching job titles against a keyword list, a method that missed roles like Business Analyst or Reporting Officer that are data jobs in substance, while letting administrative roles with "data" in the title through. As Data Quality & Insights Analyst, I was asked to build a defensible basis for how listings were actually classified. I reframed the question from what a job is called to what skills it requires, building a weighted skill-scoring framework from 34 skill keywords drawn from listing descriptions, each scored by how strongly it signals genuine data work. Applied across an extract of 7,039 listings, the framework surfaced 1,134 hidden data jobs (16.1%) that title-based sourcing had missed entirely, while flagging 1,076 listings (15.3%) as noise. Roughly one in five real data roles had been invisible to the previous method, and the classification became the basis for how the platform now categorises its job database.

Turned Reconciliation Into a Continuous Control Across 1,000+ Donor-Funded Stakeholders

Tana Water Works Development Agency, a Kenyan government agency administering African Development Bank-financed water infrastructure programmes, needed a stricter standard of financial control than ordinary government spend allows: every payment traceable to an approved programme activity, audit-ready continuously rather than annually. As Accounts Assistant, I treated reconciliation as a continuous control rather than a month-end clean-up exercise, validating supplier and contractor payment documentation against approved programme budgets before posting, and applying accrual-based recognition consistently so expenditure landed in the period it belonged to. I reconciled high-volume transactions across more than 1,000 contractors, suppliers and stakeholders, operated the agency ERP to post journals and process payments, and worked directly with procurement and programme teams because the answer to most reconciling items sat with the people who had commissioned the work, not in the ledger. Within four weeks, an investigation into expenditure trends contributed to a 9% reduction in discretionary operational spending, and financial records remained audit-ready throughout, with expenditure across donor-funded programmes substantiable against approved budgets on request.

Built a Loyalty KPI Dashboard for Woolworths Sydney Stores

As part of my data analytics training at The Data Analytics Institute, I built a Power BI KPI dashboard tracking Woolworths Everyday Rewards loyalty performance across Sydney stores. The project required translating raw loyalty and sales data into store-level KPIs that a retail operations team could act on directly. I designed the data model, the KPI logic and the dashboard layout to surface loyalty engagement and trading performance trends by store. The result is a dashboard framework directly applicable to the retail operations and corporate structure roles I am now targeting.

Academic Project · La Trobe University

Built a Credit Risk Classification Model Using Machine Learning

For my Master of Financial Analysis at La Trobe University, I built a credit risk classification model using logistic regression and random forest techniques to predict borrower risk. The project required cleaning and preparing financial and credit data, selecting and testing model features, and evaluating classification performance against real-world credit risk criteria. It sits alongside a related project on robo-advising and algorithmic trading, both applying predictive analytics directly to financial decision-making. Together, they demonstrate the data science depth behind the forecasting and dashboard work I deliver commercially.