Data Science student · Aspiring Machine Learning Engineer
Final-year BSc Data Science student · Nairobi, Kenya
$ whoami
Tony Kenga · Data Science student in Nairobi, heading into ML engineering.
$ cat mission.txt
A model in a notebook is a hypothesis.
A model people can use is an answer.
I build the second kind.
$ cat principles.txt
01 Start with the question, not the algorithm.
02 Make it reproducible, or it didn't happen.
03 Ship it, then let real use teach me what to fix.
$ status
now finishing TechPulse (final-year project)
next ML engineering internship
org Skynet DataGrid Labs
Final-year project. An ML platform that predicts a developer's technology trajectory from skill and activity signals.
Python · scikit-learn · predictive modeling
An end-to-end churn pipeline covering data, training and evaluation, built and run entirely on GitHub Actions, with no local execution.
Python · scikit-learn · GitHub Actions · CI/CD
Image classification in TensorFlow, packaged in a reproducible Docker training environment with an optimized image and containerized file management.
TensorFlow · Docker · computer vision
An independent product line of production-ready dashboards, including builds on Kenyan datasets (47 counties, FMCG sales), sold through Selar and a Google Sites storefront.
Tableau · data visualization · analytics
Scraping with rvest, plus an interactive 3D terrain visualization in Plotly.
R · rvest · Plotly
More at Skynet DataGrid Labs
| Focus | Tools |
|---|---|
| Daily use | Python · pandas · NumPy · scikit-learn · TensorFlow · SQL (PostgreSQL) · Git · GitHub Actions · Tableau · Plotly |
| Working knowledge | Docker · PyTorch · Keras · R · Power BI |
| Learning next | MLflow · Kubernetes |
Result: Second place on the public leaderboard of the Getting Started competition · RMSLE 0.00044.
Method: Lasso (0.60) and XGBoost (0.40), pruned to 77 features using an alpha sweep. Both models were averaged across five random seeds for variance reduction; XGBoost was retuned for the 77-feature set. Out-of-fold RMSE: 0.1082 using price-stratified five-fold cross-validation.
View the competition leaderboard
╔══════════════════════════════════════╗
║ ★ W A N T E D ★ ║
║ Tony Kenga · Data Science Student ║
║ Final year · Nairobi ║
║ Reward: one ML internship ║
╚══════════════════════════════════════╝
