Automated Electricity Data ETL
Engineered a Python/Pandas automated ETL pipeline for electricity sales data, eliminating manual retrieval and cutting workload by 5 days per quarter. Extracted and flattened nested JSON into optimized Parquet formats.
Building scalable infrastructure, optimizing complex ETL pipelines, and engineering backend systems that drive business intelligence.
I am a Computer Science student at the University of Saskatchewan specializing in Data Engineering and Systems Architecture. My work spans building secure backend systems for stealth SaaS startups, architecting Medallion data lakehouses on Databricks, and automating enterprise ETL pipelines.
I thrive on turning raw, messy data into high-fidelity, analytics-ready models that stakeholders can actually use to drive decisions. Whether it's standing up cloud infrastructure with Terraform, optimizing SQL queries, or orchestrating multi-agent AI systems, I build for scale and reliability.
Engineered a Python/Pandas automated ETL pipeline for electricity sales data, eliminating manual retrieval and cutting workload by 5 days per quarter. Extracted and flattened nested JSON into optimized Parquet formats.
Engineered a Medallion-architecture lakehouse on Databricks consolidating multi-country retail data across 11,000+ stores, delivering analytics-ready models for downstream Power BI insights.
Engineered an end-to-end ELT pipeline ingesting 500+ artists into a GCP Medallion architecture. Built dbt Core models served via Looker Studio. Provisioned infrastructure entirely as code using Terraform.
Production-style medallion pipeline on Databricks processing the TPC-H benchmark dataset (8 tables, ~43M rows) via serverless compute. Handled SCD Type 1 and automated data-quality checks.
Engineered a multi-agent orchestration system utilizing external LLM APIs, designing modular Python workflows with strict state management to solve complex, multi-step queries and reduce hallucinations.
Developed an AI-powered safety assistant web app utilizing Gemini 3 Pro Vision to identify and divert hazardous household e-waste. Integrated Google Maps Grounding for hyper-local safety checks.
Developed a full-stack, real-time collaborative web application using Next.js and a FastAPI backend. Engineered custom WebSocket integrations to broadcast live events across connected clients.
Built a responsive, dark-mode web application enabling rock climbers to log bouldering sessions, track send statistics, and manage active projects with custom React hooks and REST APIs.
Engineered a specialized AI tutoring system utilizing MedGemma and MedSigLIP models to process and analyze medical imaging data, designed to assist in radiology training workflows.
Engineered an ARIMA time series forecasting model to predict half-hourly electricity demand for the Australian energy market, delivering a 15% RMSE baseline.
Deep learning model built in PyTorch analyzing and predicting real-world traffic patterns utilizing LSTM architecture and machine learning pipelines.
Conducted comprehensive exploratory data analysis (EDA) on marketing campaign datasets to evaluate customer purchasing behaviors and identify high-value demographic segments.
Developed a comprehensive COVID-19 data dashboard using advanced SQL techniques and Tableau, enabling multi-metric regional analysis and revealing pandemic impact trends.
Developed an interactive Tableau executive dashboard analyzing sales and operational data to identify top-selling products, shipping cost impacts, and profit margins.