Chief Strategy Officer · Executive & Board Advisor · Digital & ICT Transformation
Credit Risk Analytics · FinTech · Forensic Accounting Systems
Cairo, Egypt
Senior executive and board-level advisor with 25+ years of leadership across ICT, FinTech, credit risk and digital transformation — at Alcatel, Nokia, Huawei, and CRIF / Dun & Bradstreet, then as Chief Strategy Officer and Chairman's Advisor to SMEs.
The work has consistently been the same problem in different clothing: taking a technically complex system and turning it into a decision a board can defend. Capital budgeting cases at Huawei, credit-scoring models at CRIF, corporate strategy as a CSO.
Now I build the analytics instruments I spent two decades buying.
| Product | What it does | Status |
|---|---|---|
| AAAS — Accrual Anomaly Audit Screening | Scores filed U.S. public-company annual statements for accrual-anomaly and earnings-manipulation risk. Six classical forensic layers — Beneish M-Score, Altman Z-Score, Dechow F-Score, Jones discretionary accruals, Roychowdhury REM, SOM clustering — plus a machine-learning ensemble trained on SEC enforcement actions. One 0–100 score across five risk tiers. | Free during early access |
| CapprossBins | Automated Weight-of-Evidence and Information-Value binning for credit scorecards. Monotonic bad-rate enforcement, special-value isolation, auto-optimal IV search, manual override. | Live |
Both are built on one principle: a score that cannot be explained cannot be trusted. Every number AAAS produces carries a show-the-math audit trail from filed line item → transformation → layer verdict → composite score, and missing data is disclosed rather than silently imputed.
- Trained and calibrated on 207,457 company-years
- Including 938 company-years under confirmed SEC enforcement
- Published case study: Archer-Daniels-Midland FY2019 — including the years the screen did not flag
AAAS produces screening signals, not verdicts. It is not investment advice, not an audit opinion, and not an assertion that any company has committed fraud.
The engines are commercial and closed-source. The public repositories here carry documentation, licensing and the public record:
- Accrual-Anomaly-Audit-Screening — methodology, layer breakdown, licence
- CapprossBins — binning tool documentation
- Live screen on Hugging Face Spaces — no account, no signup
| Period | Role |
|---|---|
| Jun 2023 – Present | Business Consultant · Executive & Board Advisor (Freelance) — corporate strategy, digital transformation and investment decisions for boards and executive teams |
| Jan 2022 – May 2023 | CRIF / Dun & Bradstreet — Global Data Analytics Projects Lead, Credit Risk. AI/ML credit scoring and decision-support models for financial institutions and credit bureaus |
| Jun 2015 – Dec 2021 | Chief Strategy Officer & Chairman's Advisor — SMEs. Board-level strategy, P&L accountability, organisation-wide transformation |
| Nov 2011 – May 2015 | Huawei — Regional Director, Professional Services (North Africa). Translated complex technical programmes into financial business cases for executive decision-making |
| May 2001 – Mar 2011 | Nokia Siemens Networks / Nokia Networks — senior global roles in business development, solution sales, network operations and service transition |
| Jul 1996 – Apr 2001 | Alcatel — Team Leader, fixed networks; promoted to Alcatel Certified Instructor by Alcatel France |
Executive competencies — Corporate & Board Strategy · P&L and Financial Governance · Digital & Business Transformation · Investment and Capital Budgeting (ROI, NPV, IRR) · Credit Risk & Analytics Strategy · ICT and FinTech Leadership · Portfolio & Performance Management · Executive Advisory
| Domain | Applied to | Tools |
|---|---|---|
| Forensic accounting analytics | AAAS scoring engine — six-layer forensic ensemble over SEC filing data | Beneish M-Score Altman Z Dechow F Jones DA Roychowdhury REM SOM clustering |
| Credit risk modelling | Scorecard development, binning, decisioning at CRIF and in CapprossBins | Weight of Evidence Information Value Logistic Regression Monotonic Binning |
| Machine learning | Enforcement-trained classification ensembles, out-of-time validation | scikit-learn XGBoost TensorFlow PyTorch |
| Data engineering & analysis | SEC XBRL panel construction, feature pipelines, model diagnostics | Python pandas NumPy Jupyter Anaconda SQL |
| Applied AI | Narrative and disclosure analysis layers, local model workflows | LangChain Ollama LLM integration |
| Delivery | Shipping analytical tools as usable products | Streamlit Hugging Face Spaces Google Cloud Platform |
| Executive & financial | Business cases, capital allocation, board reporting | ROI / NPV / IRR Financial Modelling Portfolio Management |
- From Score to Evidence: A Defensible Way to Screen Accrual-Anomaly Risk at Scale — why a score that cannot be explained cannot be trusted · also on Medium
MBA — New Orleans, Louisiana, USA
Project Management Diploma — University of Cambridge
BSc, Communications & Electronics Engineering — Ain Shams University, Cairo
Languages — Arabic (native) · English (fluent)
Open to board advisory, executive consulting, and conversations with auditors, credit teams and forensic investigators using AAAS.