Evaluating model robustness using adversarial examples and white-box attack techniques. Implements FGSM and PGD attacks to demonstrate neural network vulnerabilities and evaluate defense strategies against adversarial perturbations.
Fine-tuning large language models using Low-Rank Adaptation (LoRA) on downstream NLP tasks. Achieves strong performance with a fraction of trainable parameters compared to full fine-tuning, ideal for resource-constrained environments.
Implementing ResNet architectures for image classification on the CIFAR-10 benchmark dataset. Explores how residual connections improve gradient flow, training stability, and classification accuracy on diverse image categories.
Machine learning pipeline for predicting weather-related disasters using historical datasets. Covers end-to-end data preprocessing, feature engineering, model training, and evaluation across multiple classifiers including Random Forest and Gradient Boosting.
A Streamlit web application that predicts student exam outcomes using trained machine learning models. Users input student data and receive real-time predictions, demonstrating applied ML in an accessible, interactive format.
A marketplace mobile application where users can buy, sell, rent, or exchange goods and services. Built with React Native and SQL, supporting real transactions across Uganda with a clean cross-platform mobile experience.
A clean, dark-themed personal portfolio template for GitHub Pages. No frameworks, no build tools - pure HTML, CSS, and JavaScript. Easy to customize with a fully responsive layout, smooth interactions, and optimized SEO.
Building low-code AI agents and voice agents in n8n v2.0. Integrates ElevenLabs for voice synthesis, Agentic RAG for knowledge retrieval, and MCP for multi-context processing in production AI automation workflows.
A comprehensive collection of AI agent implementations using CrewAI, LangGraph, AutoGen, and the OpenAI SDK. Covers multi-agent systems, coder agents in Docker, and MCP integrations across multiple frameworks.
An AI-powered voice concierge built in Python. Handles natural language voice interactions by combining speech recognition and synthesis to deliver a conversational AI assistant experience.
An agentic AI system that automates sponsorship outreach using the OpenAI SDK for intelligence, SendGrid for email delivery, and Asyncio for concurrent operations. Demonstrates production-ready AI agent design patterns.
A CrewAI multi-agent system that generates one-page analyst reports in the style of top consulting firms including Deloitte, EY, KPMG, and PwC. Showcases AI-powered business intelligence workflows for professional reporting.
An intelligent job search agent powered by OpenAI's web search tool. Automates job discovery, filters relevant opportunities based on criteria, and delivers structured summaries to streamline the application process.
Hands-on exploration of machine learning systems engineering and operations. Covers model deployment, monitoring, and production ML pipeline best practices using modern MLOps tooling and workflows.
Python-based browser automation scripts for web scraping, testing, and workflow automation. Programmatically interacts with web applications to handle repetitive tasks and data extraction at scale.
Applying adversarial attack techniques to the MNIST handwritten digit dataset using PyTorch. Demonstrates how neural networks are vulnerable to carefully crafted perturbations and explores defense mechanisms against adversarial examples.
A tracking system for monitoring Covid-19 case data. Aggregates and visualizes pandemic statistics to provide insights into spread patterns, trends, and geographic distribution.
No projects match that search.