what I do
Python, Pandas, Numpy, Matplotlib, Seaborn, Plotly, Tableau, PowerBI, Streamlit
Scikit-learn, Tensorflow, Keras.
Data Structures and Algorithms, HTML, CSS, Flask, Postgresql, SQLite, Docker, Sendgrid API, Twilio API, FastAPI and GraphQL
Git and GitHub, Jupyter Notebook, Cloud Computing with Heroku and Amazon Web Services, Technical Writing, Technical Presentation.
featured work
Built a machine learning model of about 90% accuracy on transactional data to predict loan repayment.
Developed a loan repayment predictive model with selfies using Convolutional Neural Network(CNN).
Set up model deployment pipeline on Heroku and AWS.
Monitor Machine Learning models in production.
Built an ad-hoc-email sender with Sendgrid API and Postgres SQL, for firing emails automatically to about 12,000 users for the purpose of retargeting and remarketing.
Built a Deep Convolutional Neural Network model and a web app that screens microscopic images so as to detect cancer, thus increasing speed, accuracy in cancer diagnosis, and testing.
Handed over to doctors in Tanzaniaʼs labs for tests and usage.
Link
2 in every 5 women have experienced different forms of online harassment, my team and I built a working solution called 'SafeCyber' to help tackle this through allies.
Victim or friend of victim reports a case of harassment by filling the online form.
Built a classifier model of about 80% accuracy, predicts and categorizes the form data: into harassment types and action to be taken.
Form data and predictions are saved in the admin database. Records are also forwarded to the decision-makers within the ally organization.
This has helped eliminate physical and emotional stress in about 1000 women.
Link
Developed an approach by working on the Wikipedia data dump, API pages, and Protection Pages. Used Python packages in decompressing and extracting data from the dump. Did statistical analysis and sophisticated visualizations to generate insights from the data. Moreso, stacked ml algorithms were used to predict if a protection type can be edited or moved and we achieved about 75% accuracy.
Link
featured writing
Set up your Machine Learning Models for Deployment
This article is dedicated to those that’d like to learn how to build end-to-end data science and machine learning projects.
First Steps into Building your Data Science Portfolio Website
Data scientists complain about how uneasy it is to secure a data science job.
Convolutional Neural Network for Detecting Cancer Tumors in Microscopic Images
See how we built Elixir, an AI-Based Diagnosis System.
Build Quick Models with DABL
A tutorial on how to clean, preprocess, visualize and build machine learning models in three lines of code.
Get your Flask App Deployed in a Swift
Heroku shows its face where it matters. Here is a step by step tutorial on how to deploy your flask application on Heroku
Struggling to Annex Google Colab in Building your Machine Learning Models?
Remarkably, it cost an arm and a leg to get a decent GPU from existing cloud providers.