Kaisha Melckzedek Kirya
@MelckykaishaData Scientist · AI/ML Engineer · Developer ·Applied Researcher
Language Breakdown
Lines of code distribution across 26 owned repositories
I-Shaped Developer
I-shapedSpecialist — deep expertise in Jupyter Notebook
Collaboration Network
Global Impact visualization
Repos
26
PRs
0
Growth
+18%
Top Collaborators
No collaborator data yet.
Coding Streak
Contribution activity over the past year
mavo80
@mavo80
marymaina245
@marymaina245
Melbride
@Melbride
Kristine018
@Kristine018
irenee-k
@irenee-k
Top Repositories
Interactive demonstration of synthetic data generation using GANs and VAEs with statistical comparison
Real-time AI-powered emergency reporting system for Kenya. Citizens geo-tag incidents, an AI risk engine scores severity, and responders get live WebSocket alerts via a command dashboard. Built with NestJS, Next.js, React-Leaflet, Prisma, and Supabase.
A collaborative full-stack To-Do list app using HTML, CSS, JavaScript, Express.js, and Firestore.
🧠 Facial Expression Recognition CNN Model This project implements a Convolutional Neural Network (CNN) for automatic facial expression recognition using the FER-2013 dataset. The model classifies facial images into seven human emotion categories: Angry, Disgust, Fear, Happy, Sad, Surprise, and Neutral.
A collaborative mini-project that implements a **content-based recommender system** for movies. The system uses **TF-IDF vectorization** (and optionally embeddings) to build item profiles from movie metadata (e.g., plot summaries, genres, or tags) and recommends similar movies based on **cosine similarity**.
A hybrid movie recommender system built on the MovieLens 100K dataset that combines Content-Based Filtering (TF-IDF cosine similarity) and Collaborative Filtering (SVD matrix factorization) into a weighted ensemble model. The project spans a full ML pipeline — from EDA and model training in a Google Colab notebook to a dark-themed Streamlit web app
Deep learning system for hyper-local weather forecasting across all 47 Kenya counties. Trains and compares LSTM, GRU & ConvLSTM models on 10 years of historical meteorological data. Interactive 7-day forecast map built with Streamlit & Folium.
CNN-based deep learning model for classifying emotions from text into 6 categories (joy, sadness, anger, fear, surprise, love). Built with TensorFlow & Streamlit. Final year BSc Data Science project — Meru University of Science and Technology.
Open Source Impact
Contributions to external projects
No external contributions found.