ML Engineer | 6 Live Projects in Production | Physics and Mathematics | Kisumu, Kenya | Open to Remote Roles Globally
Self-taught ML Engineer from Kisumu, Kenya — building production-ready ML systems that solve real African problems, from flood prediction to salary transparency.
With a B.Sc. in Physics and Mathematics from Moi University and hands-on geophysical field research at KenGen Olkaria Geothermal Project, I bring scientific rigour to every model I build. Every project ships with a live URL, SHAP explainability, and a real problem it solves — built entirely on Android (Termux, Google Colab).
- B.Sc. Physics (Major) + Mathematics (Minor) — Moi University, 2012
- Industrial Attachment — KenGen Olkaria Geothermal Project, 2011
- Research — Eburru Geothermal Prospect geophysical study
- Location — Kisumu, Kenya (near Lake Victoria)
- Timezone — EAT (UTC+3), Available 08:00-17:00, Async-friendly for EU/US
- Open To — Remote ML Engineering, AI Engineering, Data Science roles globally
- Exploring: LLM Engineering — RAG pipelines, pgvector on Neon, LangGraph agentic AI
- Building: Sokoni — WhatsApp AI sales agent for Kenyan SMEs (in progress)
- Next: QLoRA fine-tuning on Llama 3 and Mistral via Colab T4
- Ask me about: flood prediction, African salary data, church management SaaS, K-Means segmentation
- Build end-to-end ML systems — raw data to feature engineering to model training to live API
- Apply Physics and Mathematics background to feature engineering and model evaluation
- Reduce real disaster risk — Nyando Flood AI gives 161,000 Kano Plains residents earlier warning
- Ship production-grade code — FastAPI, Flask, Docker, GitHub Actions CI/CD
- Explain every prediction with SHAP — because unexplainable AI is not good enough
- Build secure multi-tenant SaaS — JWT rotation, RBAC, M-Pesa payment integration
- Localise global ML techniques to African business contexts — M-Pesa, Safaricom, BNPL
Languages and Data
Machine Learning and AI
Web and Deployment
| Project | Description | Stack | Live |
|---|---|---|---|
| Nyando Flood AI | GradientBoosting, 2308 GEE satellite points, AUC 0.9717, F1 0.9022, 161,000+ residents, 41 CI tests, Zenodo | scikit-learn, FastAPI, Docker, React | API |
| ChurchOS | Africa-first multi-tenant church SaaS, M-Pesa and Flutterwave, JWT auth, 5-role RBAC, Finance audit log | Flask, React, PostgreSQL, Render, Vercel | Demo |
| AfriSalaries | XGBoost salary band classifier, 8 African countries, E2E 88% accuracy, HIGH precision 0.72, 1526 real rows | XGBoost, FastAPI, Docker, React, Vercel | App |
| SegmentIQ | Live K-Means ML app, customer segmentation from age, income, spending score, CSV batch upload, 5 segments | scikit-learn, FastAPI, React, Render, Vercel | App |
| Loan Risk Assessment | Basel III framing, Gini 0.74, IFRS 9 staging, EL = PD x LGD x EAD, saves 23% cost | scikit-learn, pandas, FastAPI | — |
| Titanic Survival | Leak-free Pipeline, SHAP waterfalls, StratifiedKFold, Bootstrap CIs, Zenodo, AUC 0.8661 | scikit-learn, SHAP, FastAPI | Repo |
[
](https://github.com/jameskoero)
[
](https://github.com/jameskoero)
| Status | Project | Domain |
|---|---|---|
| Live | Nyando Flood Risk AI | Climate / Disaster |
| Live | ChurchOS | SaaS / Web App |
| Live | AfriSalaries | Labour Economics |
| Live | SegmentIQ | ML / Customer Analytics |
| Live | Loan Risk Assessment | FinTech / Banking |
| Live | Titanic Survival | Education / Portfolio |
| In Progress | Sokoni — WhatsApp AI Sales Agent | Agentic AI / SME Commerce |
| Learning | RAG Pipeline — LangChain + pgvector + Neon | LLM Engineering |
| Learning | LangGraph Multi-Agent Systems | Agentic AI |
| Planned | QLoRA Fine-tuning on Colab | LLM Engineering |
| Certificate | Issuer | Date | Credential |
|---|---|---|---|
| Machine Learning using Python | Programming Hub and Google Developers Launchpad | Oct 2025 | bae4cf502b3dfe5 |
| Python Basics | Programiz | Sep 2025 | 08ddece2-fd4c-40eb-88d9-8f6b142466b0 |
B.Sc. Physics (Major) + Mathematics (Minor) — Moi University, Kenya 2008-2012
Classical Mechanics, Statistical Physics, Linear Algebra, Calculus, Numerical Methods
Research: Eburru Geothermal Prospect (MT and TEM methods)
Industrial Attachment — KenGen Olkaria Geothermal Project 2011
Large-scale geophysical survey data collection at the most productive geothermal field in Africa. Applied MT and TEM subsurface imaging — first exposure to scientific data pipelines at production scale.
I came into ML from geophysics — where a wrong model meant a wrong drill decision worth millions of shillings. That background is why every model I ship has a number attached to it, not just accuracy, but what the number means in the real world. Every system I deployed started on a 6-inch screen with 4G data from Kisumu, Kenya. Constraints sharpen thinking.
Available for remote ML Engineering, AI Engineering, Data Science roles globally.
Building from Kisumu, Kenya — one model at a time.