Workflows
End-to-end computational workflows implemented as fully-documented Jupyter notebooks — covering real satellite and scientific text datasets from raw acquisition through validated final outputs.
Satellite Remote Sensing
-
NASA PACE Ocean Color Analysis
Atmospheric correction and spectral analysis of PACE OCI hyperspectral data targeting phytoplankton community classification and ocean colour retrieval from continuous-band sensor measurements.
-
Preprocessing, spectral index computation, and classification applied to Sentinel-2 time-series imagery for land cover change detection mapping.
Applied NLP for Scientific Text
-
Transformer pipelines for extracting domain entities, geographic markers, and scientific terminology from unstructured research text.
-
Topic Modeling & Semantic Structuring
Organising unstructured scientific abstracts into latent thematic clusters using LDA and neural topic models.
-
Zero-Shot Document Classification
Pre-trained language models for zero-shot tagging and classification of research documents on arbitrary criteria without task-specific fine-tuning.
Running These Notebooks
The raw .ipynb source for every workflow lives in the site repository, under
docs/tutorials/.
Download a single notebook from there, or clone the lot:
The pixi environment in this repository builds the site; it does not install Jupyter or
the scientific stack each notebook needs. Run them under your own environment, installing
the imports each notebook declares in its first cell.