Every project in pipeline format. Input, process, output. Grouped by signal type. The full chain from source to delivery.
Built a full scraping and intelligence system that monitors Reddit job communities in real time, classifies posts using NLP, and serves insights through a live dashboard. Designed to cut through noise and surface actionable job market signals.
High-concurrency async scraper targeting Sydney commercial property developers. Extracted structured lead data from DA PDFs and leasing brochures across multiple council portals, delivering clean prospect lists for a real estate client.
Targeted lead list for a concrete cutting company expanding into Melbourne growth corridors. Extracted verified contacts from construction directories with strict accuracy requirements on ABN, service area, and contact validity.
Compiled a targeted email list of Swedish metal and steel manufacturers for a B2B outreach campaign. Extracted company profiles, decision-maker contacts, and verified emails from European industrial directories.
End-to-end system: scraped 10,000+ rental listings from a JS-rendered Nigerian property platform, engineered location and property features, trained a Random Forest model, and deployed predictions via a Flask API.
Validated 480+ ESG controversy records for Pakistan Stock Exchange listed firms. Built an LLM-powered pipeline that cross-referenced claims against source data, flagged inaccuracies, and produced a fully traceable correction log.
Built a fully automated content pipeline that generates, schedules, and publishes posts to X (Twitter). Uses Claude API for content generation, Make for orchestration, and Buffer for scheduling. Runs hands-free.
Analyzed 547,918 POS transactions ($6.2M revenue) for a restaurant chain using menu engineering methodology. Classified every menu item as Star, Plowhorse, Puzzle, or Dog. Identified $209K-$271K in annual profit improvement opportunities.
Built an interactive dashboard for a Shopify store processing 65,000+ transactions ($2.69M revenue). Covers KPI tracking, seasonal trends, payment method analysis, and heatmap visualizations for sales patterns.
Full ML pipeline for predicting customer churn in a telecom dataset. Automated preprocessing, feature engineering, and model training with Random Forest and Logistic Regression. Evaluated with accuracy, precision, recall, F1, and ROC-AUC.
The modeling layer of the Lagos Rent system. Trained a Random Forest regressor on engineered features from 10,000+ scraped listings. Deployed as a Flask API for real-time price predictions based on location, size, and property type.
Spatial analysis of heavy metal concentrations in groundwater samples. Mapped contamination hotspots using GIS tools, assessed health risk indices, and produced visualizations for environmental compliance reporting.
Interactive dashboard exploring the relationship between electricity consumption and GDP across countries. Visualizes development patterns, energy intensity trends, and regional comparisons.