Proof of Work
Portfolio & Case Studies
Real projects. Real results. Here's what's possible when data expertise meets execution.
📋
Power BIDAX
Power AutomateOneDriveExcel
BPSS Project Management Dashboard
Confidential Client · India
Designed and delivered a fully automated 4-dashboard Power BI ecosystem for a confidential client — tracking 100 projects across 5 global regions (AMER 17%, APAC 19%, EMEA 24%, LATAM 22%, NORAM 18%) and 6 Project Managers. The entire pipeline is automated: Excel on OneDrive is refreshed by a Power Automate flow, feeding live data into Power BI with zero manual intervention. Of the 100 projects, 26 are completed (26% completion rate), 33 delayed, 22 pending, and 19 in progress — with turnaround times ranging from 87 to 562 days.
📊Executive Overview
KPIs, regional donut chart, monthly trend bar (Jan 2023–Jul 2026), status breakdown (Completed/Delayed/In Progress/Pending) — all filterable by date, PM, and region.
🎯Project & PM Tracker
Granular project table with Region, Month, Status, Avg Rating & Turnaround Days. Workload chart for all 6 PMs (Casey Chen & Taylor Morgan lead at 21 projects each, Sam Wilson lowest at 11).
🔬Quality Deep Dive
Heatmap rating matrix across 9 criteria per project (avg 2.92–3.11). Avg ratings by PM (Alex Rivera highest at 3.64) and Region (EMEA highest at 3.54, LATAM lowest at 3.32) with conditional colour coding.
📈Project Trends
Turnaround trend line, stacked experience bar by month, scatter plot of Turnaround vs Avg Rating by region, plus 3-year YoY comparison (2023: 3.51–4.61 → 2025: 3.76–4.74).
▸Configured end-to-end Power Automate pipeline — zero manual data entry required
▸Advanced DAX measures for YTD scoring, rating bands, completion rates, and turnaround KPIs
▸Integrated Power BI Q&A natural language visual for stakeholder self-service queries
▸Cross-page synced slicers — filter by date, status, PM, and region simultaneously
▸Identified PM workload imbalance and regional quality variance — enabling targeted coaching and resourcing decisions
📋 100 Projects🌍 5 Global Regions
⚡ Fully Automated📐 4-Page Report🗣️ NL Q&A Visual📅 3-Year Trend Analysis
🏥
Power BI
DAX
Power Query
Star Schema
ETL
Hospital Procurement & Supply Chain Analytics
Healthcare Network · Nigeria
Comprehensive 4-page Power BI dashboard for a healthcare network's procurement team — replacing manual Excel reporting with real-time interactive insights across 92 hospitals, 611 medical equipment units, and 259 equipment types. Built a full Power Query ETL pipeline transforming raw procurement data into a clean star schema, surfacing top cost drivers and supplier performance for the first time.
📊Overview
Total spend ($20.87M), cost by equipment type, delivery trend by date ordered, cost by location — filterable by department, hospital, state, and delivery date.
🔧Equipment Analysis
Lifecycle condition breakdown (53.16% new, 25.32% refurbished, 21.52% used-good), quantity by department (Pediatrics 164, Radiology 139, Emergency Care 112), full cost table across 15+ equipment types.
🚚Supplier Performance
Delivery variance analysis across 7 geographic locations, vendor reliability ranking, average delivery time of 25.31 days benchmarked across vendors.
🌍Geographical Insights
Regional spend distribution (Minna City $1.82M top, Yenagoa $1.43M, Dutse $1.34M), city-level procurement hotspot identification.
▸Engineered end-to-end ETL pipeline in Power Query — star schema across hospital, equipment, supplier and date dimensions
▸Identified top cost drivers: MRI Machines $5.2M, CT Scanners $4.7M, Operating Tables $2.1M — enabling vendor negotiation
▸Advanced DAX measures: lifecycle cost tracking ($23M total), equipment condition segmentation, geographic cost aggregation
▸Enabled preventive maintenance planning through equipment lifecycle visibility across 611 units
💰 $21M+ Tracked
🏥 92 Hospitals
⚡ Real-Time
📐 Star Schema + Advanced DAX
🚚 Supplier Benchmarking
🏨
Power BI
DAX
Power Query
ETL
Healthcare Analytics
Executive Command Centre — Healthcare Operations
Regional Healthcare Network · Status: Completed
A regional healthcare network had no unified operational view — executives couldn't distinguish structural failures from short-term demand shocks, and reporting was retrospective and fragmented. Built a 4-dashboard executive intelligence platform unifying bed management, staffing, patient outcomes, and demand forecasting into a single trusted view, powered by 30+ DAX measures across 6 unified data sources with automated Green/Amber/Red risk thresholds on every KPI.
📊Executive Overview
Real-time KPI cards with traffic-light status (Bed Occupancy 26.9% 🟢, Readmission Rate 25.2% 🔴), bed & staffing coverage stability trends, emergency vs elective admissions, AI-generated insights & recommended actions.
🛏️Capacity & Throughput Control
LOS trend & pressure analysis (5-day risk threshold), bed utilisation vs capacity, average LOS by facility type, delayed discharge risk by facility (Hospital ~47 vs Clinic ~12).
👥Workforce & Cost Control
Staffing coverage stability (95%/90% thresholds), overtime pressure trend (70-hour risk threshold), staffing gap by facility (Hospital 1,132 vs Clinic 377 headcount), 113K active staff monitored.
📡Demand Intelligence & Early Warning
Emergency admission trend & forecast (754 admissions, 7.57 avg forecasted), readmission risk monitoring, external pressure vs admissions dual-axis chart, 7.56% demand growth rate.
▸Unified 6 data sources (locations, patients, admissions, staffing, capacity, external calendar) into a single star schema with calculated date table
▸Engineered 30+ DAX measures with automated Green/Amber/Red threshold logic on every executive KPI
▸Surfaced a 25.2% readmission rate as a red-alert KPI with automated recommended clinical interventions
▸Quantified a 2K headcount staffing gap and 5K overtime hours — enabling evidence-based workforce reallocation
▸Shifted the organisation from retrospective reporting to anticipatory governance
🏨 4-Dashboard Suite
📊 30+ DAX Measures
🟢🟡🔴 Threshold Alerts
📈 Demand Forecasting
👥 Workforce Analytics
📁
Excel
Power Query
M Language
Data Auditing
HR Data Reconciliation & Auditing System
Mr. Jimoh · Status: Completed (incl. client training)
Client was manually reconciling HR/payroll data between 2024 and 2025 employee datasets — slow, error-prone, and unable to detect ghost employees, underage employment, or suspicious salary changes. Built a one-click automated audit framework in Excel using Power Query, replacing the entire manual process with 15 systematic checks across data integrity, salary anomalies, workforce movement, and compliance — reducing processing time by 60%. Delivered full Power Query training so the client now runs and extends the system independently.
🔴Data Integrity (7 checks)
DOB mismatch across years, missing DOB (2024/2025), DOB after employment date, employment after valuation date, DOB after valuation date.
🟡Salary Anomalies (2 checks)
Salary decrease detection, salary increase >50% flag.
🟢Workforce Movement (3 checks)
Missing from 2025 records (ghost employee detection), new hires 2025, new exits 2025.
🔵Compliance & Policy (3 checks)
Underage at employment, underage at valuation, over-60 at valuation (retirement policy).
▸Built 15 automated M Language audit queries across 3 source datasets (EmployeeData_2024, EmployeeData_2025, ExitData_2025)
▸Detected ghost employees — records present in 2024 with no trace in 2025 active or exit data, impossible to catch manually at scale
▸Flagged salary anomalies and DOB/timeline integrity violations automatically on every refresh
▸Automated compliance checks for underage employment and retirement-age policy
▸Delivered full client training — one-click refresh, result interpretation, and adding new checks independently
📁 15 Audit Checks
⚡ 60% Time Saved
🔍 Ghost Employee Detection
🎓 Client Training Delivered
🔄 One-Click Refresh
🛒
PythonRandom ForestStreamlitSHAP
Supermarket Revenue Forecasting
End-to-end ML pipeline for a supermarket chain to forecast peak-demand revenue. Random Forest regression with SHAP explainability so stakeholders understood why each prediction was made.
▸Achieved 99.8% R² accuracy on holdout test data
▸SHAP values for full model explainability and business transparency
▸Deployed interactive Streamlit dashboard for real-time demand analysis
🎯 R² = 99.8%📊 Live Dashboard🔍 SHAP
🫁
TensorFlow
CNN
Transfer Learning
COVID-19 Radiography Detection
Deep learning CV model to classify chest X-rays for COVID-19 detection. Transfer Learning with pre-trained CNN architectures for maximum accuracy on limited medical data.
▸Built with TensorFlow/Keras and Transfer Learning
▸Data augmentation and fine-tuning for robust medical imaging
▸Multi-class: COVID vs. Normal vs. Viral Pneumonia
🧬 Deep Learning
🩻 X-Ray Classification
⛵
Computer Vision
CNN
Multi-Class
Venice Boat Classification
Multi-class Computer Vision model classifying boat types in Venice waterways. Custom CNN architecture optimised for fine-grained visual distinctions between vessel categories.
▸Custom CNN architecture for fine-grained classification
▸Batch normalisation and regularisation throughout
🖼️ Image Recognition
🏛️ Multi-Class CNN
