ML/AI Serving & Deployment
ML model deployment with Modal
Serving and deploying a sticker sales forecasting machine learning model using Modal.
ModalSales ForecastingDeploymentFastAI
Machine learning engineer building AI/ML systems to solve business problems.
This spans everything from defining a problem and metrics, data engineering and analysis, model building to ML predictions integrated into production software, with a retraining loop that monitors for change and keeps the system accurate over time.
I am available for freelance & contract work
Available for work
Data Engineering
Extracting, transforming, and loading data
Machine Learning
EDA, feature engineering, model building
Serving and deployment
Serving, deploying, monitoring, API integration, retraining
Software Development
Python, HTML, CSS, JavaScript, Reactjs, Next.js, FastHTML
Selected work
A selection of recent work across machine learning, data science, and web development.
Python Development
Matrix Multiplication from scratch
Doing Matrix Multiplication from scratch (using just python , its standard library, matplotlib) in Solveit.
PythonFirst PrinciplesSolveit
ML/AI Serving & Deployment
ML model deployment with BentoML
Serving and deploying a binary classification machine learning model using BentoML.
BentoMLMachine LearningDeploymentFastAI
AI in Healthcare
Building an AI medical assistant
An AI system for assisted differential diagnosis based on the Uganda Clinical Guidelines.
AI in HealthcareMachine LearningMedical-LLM
Machine Learning
Stable Diffusion from the Foundations
Comprehensive walkthrough of Stable Diffusion, starting from the theoretical foundations and building up to a complete implementation.
Stable DiffusionGenerative AIAI Art
Machine Learning
Machine Learning
Machine Learning
Binary Prediction of Poisonous Mushrooms
A classification model to predict whether a mushroom is edible or poisonous from its physical characteristics.
Binary ClassificationFastAIGradient BoostingBiology
Machine Learning
Binary Classification of Insurance Cross Selling
A classification model to predict which customers respond positively to an automobile insurance offer.
Binary ClassificationInsuranceNeural NetworksFastAI
Computer Vision
Artificial Intelligence
Flexible Functions
An AI lab building practical AI use cases in the most cost-effective way possible.
AIMachine LearningDataset
Web Development
Zigoti Coffee Website
A Ugandan coffee house covering the whole value chain — growing, buying, processing, roasting, and export of coffee beans.
ReactE-commerceCoffeeUganda
Coding Challenges
Advent of Code
Solving programming puzzles and challenges from the Advent of Code series.
PythonCoding ChallengesAlgorithmsPuzzles
About me
Over the years, I've worked across the full lifecycle of ML products, which includes translating business problems into performance metrics tied to business objectives, data engineering and analysis, building simple baselines before introducing machine learning, model building, serving, deployment, and integrating model predictions into software, while setting up monitoring and the retraining feedback loop needed to keep models accurate and useful over time. Depending on the project, this has meant owning that process end to end, or focusing specifically on model development, serving, and deployment within an existing pipeline.
I’ve built predictive and generative models with Python, PyTorch, fastai, sklearn and various other common algorithms. In addition, I have 9+ years building software applications with HTML, CSS, JavaScript, React, Next.js, and other modern web technologies.
7+ yrs
Machine learning
9+ yrs
Software development
Kaggle
Competitions expert
HF
Models on HuggingFace
My background
A strong foundation in software engineering, and I apply these principles while working across the machine learning lifecycle.
- Exploratory Data Analysis and Data Wrangling
- Feature Engineering
- Model Building and Evaluation
- Deployment, Testing, and Monitoring