Camera Trap Detection
Transfer-learning pipeline for wildlife camera-trap imagery. Exploring fine-tuning of pretrained vision models against a small, noisy field dataset.
02 / Work
The analytics engineering catalog: real business segmentation work, end-to-end pipelines, dashboards, ML projects, and certifications.
The Target: Catch a fraud transaction at high precision in as little as 5 milliseconds after the purchase attempt is made. The tech: Azure Event Hubs, Databricks, SQL, Python, and Power BI.
Recency-Frequency-Monetary segmentation built on real guest transaction data, producing customer groupings the business can target — delivered in Power BI.
→ Most Recently Completed Project · AIA fully functional e-commerce store representing a sandbox for 3 n8n agents to price, stock, and purchase zoo tickets in real time. I analyzed the bot performances to identify bottlenecks in performance, and updated them to run more efficiently.
→ Modeling · EnsembleEnsemble learning predicts churn on validation data with 93% accuracy — viable for real-world business cases. Next steps: generalization tests and deployment strategy.
Transfer-learning pipeline for wildlife camera-trap imagery. Exploring fine-tuning of pretrained vision models against a small, noisy field dataset.
Exploring medallion architecture with Microsoft Fabric — bronze ingestion from Supabase, silver cleansing, gold dimensional models powering the Owl Park Power BI dashboards.
↗ Certification · MicrosoftMicrosoft Certified: Power BI Data Analyst Associate — semantic modeling, DAX, visualization, and report deployment.
↗ CertificationSnowflake Data Warehousing Workshop — strengthening cloud warehouse modeling and SQL workflow patterns.
↗ CertificationLearnSQL.com Professional Certificate of Competency, emphasizing query writing and analytical problem solving.
Hands-on Microsoft Fabric workshop covering OneLake, lakehouses, warehouses, and Real-Time Intelligence end-to-end.
A fully working ecommerce store for purchasing pet services such as dog walking.