Work
From data to insight.
Every project on this page is labelled with what it is. Client work is only published with the client’s permission.
- Fig. 1
Job costs updating through the day
View full size ↗ (opens in a new tab)Source: demo dashboard, fictional data Client project · Victoria, BCJob-costing dashboard for a restoration contractor
The challenge
The crew already logged hours and photographed receipts in their time-tracking app, but nothing connected that information to individual jobs. Knowing whether a job was profitable meant hours of sorting after the fact.
What I built
An automated pipeline that collects the crew’s hours and photos through the workday. AI reads each photo, separates receipts from progress shots, and adds materials and labour to the right job. A secure web dashboard shows the results by job and by week, with every receipt image one click away.
Shaped by the client’s needs
Each feature came from a real situation. Refunds are detected and subtracted automatically. Archived jobs with new activity are flagged for review. Misfiled receipts move to the right job in one click. Estimates sit next to actual costs. Weekly cost totals match the client’s weekly pay cycle.
The result
The crew didn’t change how they work. The owner now uses the dashboard day to day.
Tools
- Python
- Google Cloud
- Firebase
- AI vision
- QuickBooks Time
- QuickBooks Online API
- Fig. 2
Any institution’s trend next to the BC benchmark
View full size ↗ (opens in a new tab)Source: Statistics Canada, Table 37-10-0277-01 Portfolio projectPublic dataBC post-secondary enrolment dashboard
Enrolment figures for BC institutions exist in Statistics Canada tables, but comparing one institution against the provincial picture takes hours of manual work. Using Table 37-10-0277-01 (2013/14–2023/24), I loaded the data into BigQuery, built Dataform pipelines with separate development and production environments, and modelled it in Power BI as a star schema with DAX measures for enrolment, growth, market share and student profile. The result is an interactive dashboard where any institution’s trend sits next to the BC benchmark.
Tools
- BigQuery
- Dataform
- Power BI
- DAX
- Fig. 3
Net total directional connectedness over time
View full size ↗ (opens in a new tab)Source: Fig. B2, Appendix B (TVP-VAR, US context), Gaies, Chaabane, Adeosun & Sahut (2025), Energy Economics Dynamic Interactions between Climate Transition Risk, ESG Sentiment, and Stock Markets
How much do climate-transition risk and ESG sentiment move market value, and when? Using high-frequency European market data, I harmonized sources in SQL, ran time-series diagnostics, and applied TVP-VAR and quantile coherency models in R to trace how risk transmits between markets during periods of stress.
Tools
- SQL
- R
Citation
Gaies, B., Chaabane, N., Adeosun, O. A., & Sahut, J. M. (2025). Climate transition risks, ESG sentiment and market value: Insights from the European stock market. Energy Economics, 148, 108605. https://doi.org/10.1016/j.eneco.2025.108605
- Fig. 4
Wavelet coherence: Bitcoin and 5-year breakeven inflation (T5YIE)
View full size ↗ (opens in a new tab)Source: Gaies, Chaabane, Arfaoui & Sahut (2024), Research in International Business and Finance Crypto market stress testing: inflation and financial risk
Are Bitcoin and Ethereum a hedge against inflation, or just another risk asset? On a daily dataset built for 2018–2023 in Python, SQL and R, I ran statistical diagnostics (Jarque-Bera, BDS, Kruse, Hu & Chen) and applied wavelet coherence and quantile coherency to show how the dependence changes across calm and stressed regimes.
Tools
- Python
- SQL
- R
Citation
Gaies, B., Chaabane, N., Arfaoui, N., & Sahut, J. M. (2024). On the resilience of cryptocurrencies: A quantile-frequency analysis of Bitcoin and Ethereum reactions in times of inflation and financial instability. Research in International Business and Finance, 70(A), 102302. https://doi.org/10.1016/j.ribaf.2024.102302
What the labels mean
- Published research
- Peer-reviewed research; the badge links to the published paper.
- Client project
- Work for a named client, published with their permission.
- Anonymized project
- Real work for an organization that can’t be named.
- Portfolio project
- Self-directed work that shows a method or tool.
- Demonstration
- Built to illustrate a technique, using illustrative data.
Let’s make your data more useful.
Have a reporting, analytics or data challenge? Let’s discuss what you’re working with and what you need to accomplish.