Data Science projects
Selected projects showcasing Python workflows, machine learning implementation, and explainable outcomes.
ExoVision
1st place NASA Space Apps Hackathon Local
-
ExoVision uses artificial intelligence to analyze data from satellites like Kepler and detect exoplanets.​
-
Benchmarks six ML models and lets you switch among them in the app, showing each model’s accuracy/metrics to compare predictions side by side.
-
It helps astronomers sort through millions of signals in seconds, accelerating the search for new worlds beyond our solar system.​
-
It helps anyone learn all about exoplanets and space-related topics
​
Exovision is still under development and is publicly available at: https://exovision.streamlit.app/
Augmented AI Clustering
Capstone project
An end-to-end workflow to uncover topics in large text collections and assign human-readable labels at adjustable granularity. We embed sentences with Sentence-Transformers, project to 2D with UMAP, cluster using Agglomerative Clustering, and tune hyperparameters with Optuna (silhouette as the objective). The best run generates a dendrogram for exploration, and a local LLM (Ollama, “cal-state-words”) proposes labels for cluster intersections. Ideal for decision makers who need fast, explainable structure without pre-existing tags.
Chekc GitHub: https://github.com/HiramZ04/Cal-State-UMAP_Clustering
