BI · Data Science · Machine Learning · LLM Apps · MLOps · Data Engineering
Portfolio Hub · Showcase Deck (PDF) · Showcase Deck (PPTX) · precious.ikebude@outlook.com
I bridge executive Business Intelligence (Power BI, SQL, governed KPI semantic layers, financial and risk reporting) and production AI/ML systems (calibrated fraud scoring, credit-risk scorecards, read-only NL-to-SQL agents, and containerised MLOps pipelines on FastAPI, Airflow, MLflow, and Docker).
With a background in Economics (B.Sc., 2:1) and 3+ years delivering analytics, data governance, and machine learning systems across fintech, banking, and enterprise consulting, I focus on what happens after the notebook: reproducible pipelines, calibrated probability outputs with decision thresholds tied to business cost, and interfaces that risk and operations teams actually use.
📄 Solutions Showcase Deck (2025–2026): Browse the 26-slide architecture & product walkthrough — Download PDF | Download PowerPoint (.pptx)
| Product | Repository | Highlights & Verified Metrics | Stack |
|---|---|---|---|
| Moniguard v2 PoD & Mobile-Money Fraud Scoring |
ikebude/MLOps-Fraud-Detection |
Fraud scorer (0–1000 score + reason codes) on 60,000 synthetic Nigerian pay-on-delivery transactions with a strict 70/30 chronological split. PR-AUC 0.877 (95% CI 0.845–0.905), ROC-AUC 0.943, recall 0.892 at a 5% alert rate, 47.9 ms p95 FastAPI latency. | Python, XGBoost, Isolation Forest, FastAPI, MLflow, Airflow, Docker |
| Lolo's Ledger (FXLedger) Trading Journal, Risk Engine & Analytics |
ikebude/FX-Trading-Journal · ikebude/AI-Trading-Intelligence |
Shipped desktop journal (521 unit tests + 25 Playwright E2E tests passing) plus a Python risk and expectancy engine with behavioural tilt detection. Personal-edge attribution is gated until 100+ closed trades and never produces buy/sell signals. | TypeScript, React, Electron, NestJS, Playwright, Python |
| BI Operations Agent Governed NL-to-SQL & KPI Anomaly Copilot |
ikebude/AIMLOps-BI-Operations-Agent |
In development. Read-only NL-to-SQL over a governed semantic layer with KPI anomaly alerts. Benchmarks will be published once built. | Python, FastAPI, DuckDB |
| LoanLens & Customer Intelligence Credit Scorecard & Churn/CLV |
ikebude/ai-ml-data-engineering |
LoanLens: leak-free default model on 55,518 closed Prosper loans, tested on future loans: ROC-AUC 0.707, KS 0.305, with an approval-policy simulator. Customer Intelligence: churn PR-AUC 0.841 / ROC-AUC 0.933 on 7,043 customers, RFM segments and a weekly revenue-at-risk call list. | Python, Scikit-learn, Streamlit |
| AIOps Expense Tracker NGN-First Expense Intelligence |
ikebude/AIOps_Expense_Tracker |
In development. Rebuilding the Code with Mosh course starter with NGN bank-CSV import and an ML categoriser. | Python, FastAPI, React |
Company property of 3Consulting Limited. Presented publicly as architectural case studies with live links and stack documentation — no proprietary source code is hosted in public repositories.
| Solution | Live Production Link | Case Study | Architecture & My Contributions |
|---|---|---|---|
| SafeData NDPA / GDPR Data Retention & Disposal Platform |
safedata.3consult-ng.com |
View Case Study | Multi-tenant data-retention enforcement, Records of Processing Activities (RoPA), DSAR & consent-withdrawal disposal workflows, immutable SHA-256 audit hash-chaining, and automated PII discovery across enterprise data stores. |
| FraudGuard Real-Time Transaction & Login Fraud Scoring |
fraudguard.3consult-ng.com |
View Case Study | Hybrid rule + ML ensemble (0–1000 calibrated score + reason codes) for payment, PoD, and account-takeover events, velocity feature store, analyst triage & SAR case queue, and PSI model-drift monitoring. |
| Africa Stablecoin Intelligence Pan-African Stablecoin & Regulatory Tracker |
africastablecoinintelligence.com |
View Case Study | Live market-corridor analytics, USDT/USDC/cNGN liquidity & FX premium pipelines, 54-country central-bank regulatory intelligence tracker, and editorial research portal. |
Portfolio Hub — ikebude/ai-ml-data-engineering
All 11 Machine Learning, Statistical Modelling & Business Intelligence projects live in the unified monorepo (upgrades to the remaining projects are in progress):
- Prosper Loan Credit Risk (
prosper-loan-risk) — Out-of-time default scorecard and policy simulator (ROC-AUC 0.707, KS 0.305 on 13,880 future loans). - Waze User Churn Prediction (
waze-churn-prediction) — Behavioural churn analysis on 14,999 users. - TLC NYC Cab Fare Intelligence (
tlc-cab-fare-intelligence) — NYC Yellow Taxi trip analysis and fare modelling. - Customer Intelligence (
customer-retention) — Calibrated churn (PR-AUC 0.841), RFM segmentation and a CLV backtest. - S&P 500 Financial Clustering (
sp500-analysis) — PCA dimensionality reduction + K-Means regime clustering across S&P 500 fundamentals, valuation multiples, and volatility profiles. - Cars Model Analysis (
cars-model-analysis) — Vehicle specification and pricing analysis.
- CRM B2B Sales Funnel Analysis (
crm-sales-analysis) — Pipeline stage leakage diagnostics, rep win-rate quartiles, and cycle-time attribution across 8,800 B2B opportunities. - Executive Power BI Dashboards (
power-bi-dashboards) — Star-schema data models, DAX time-intelligence measures, and row-level security (RLS) across Finance, HR, and Sales packs. - Enterprise BI & Relational Analytics (
bi-analytics) — Data quality remediation framework, customer wealth segmentation, and supply-chain margin diagnostics. - Global Data Analytics Suite (
global-data-analytics) — Cross-market cohort retention, content catalogue clustering, and macroeconomic panel studies. - Production SQL Querying Patterns (
sql-querying) — Advanced CTEs, window functions, cohort retention queries, and revenue waterfall reconciliations.
- Business Intelligence & Analytics: Power BI (DAX, Power Query M, Star Schema, Row-Level Security), Tableau, Advanced Excel (VBA, Pivot Models), KPI Semantic Layers, Executive Storytelling
- Machine Learning & Data Science: Python (
scikit-learn,LightGBM,XGBoost,statsmodels,SHAP,pandas,numpy), Cost-Sensitive Imbalanced Classification, Probability Calibration, Credit Scorecards (WoE/IV), RFM & BG/NBD CLV, Time-Series Forecasting - AI Agents & MLOps: FastAPI, MLflow, Apache Airflow, NL-to-SQL Guardrails, DuckDB, SQLAlchemy, Docker & Docker Compose, GitHub Actions CI/CD, Azure Container Registry, Model Drift Monitoring (PSI / KS / MAD)
- Databases & Engineering: PostgreSQL, MS SQL Server, SQLite, DuckDB, TypeScript / React / Electron / NestJS
- B.Sc. Economics — University of Nigeria, Nsukka (Second Class Upper Division / 2:1, Oct 2018 – Aug 2023)
- Google Advanced Data Analytics Professional Certificate — Coursera (Apr 2025)
- Microsoft Certified: Azure AI Engineer Associate — (In View, 2026)
- Managerial Economics & Business Analysis Specialization — Gies College of Business, UIUC (Jun 2023)
- Email: precious.ikebude@outlook.com
- Portfolio Monorepo: github.com/ikebude/ai-ml-data-engineering
- Solutions Deck (PDF): Precious-Ikebude-Solutions.pdf
- Kaggle: kaggle.com/ikebude