Learn it.
Build it.
Break it a little.
The projects get more ambitious as you move through the notebook: first understanding how models behave, then working with biological data, and eventually connecting multiple ideas into larger AI systems. Projects marked planned are part of the Labs roadmap and are not yet interactive.
Medical AI Bias Lab
Can an AI look accurate overall while performing very differently across groups?
Foundations
Start with models you can see and interrogate: images, predictions, errors, and information retrieval.
Skin Disease Detection
Explore how an image classifier learns visual patterns and where confidence, errors, and dataset composition become important.
Histopathology Cancer Detection
Investigate how deep-learning systems can analyze patterns in pathology imagery — and what their predictions do and do not tell us.
Biomedical Literature Assistant
Explore retrieval and language models by finding, organizing, and explaining information from biomedical research.
Biological Systems
Move beyond images and text into biological sequences, gene expression, and molecular data.
Protein Function Prediction
Investigate how patterns in protein data can become features for computational predictions about biological function.
CRISPR Guide Lab
Explore the kinds of sequence features that can be used when computational models evaluate candidate guide RNAs.
Single-Cell RNA Explorer
Work with simplified single-cell gene-expression data to explore dimensionality, clusters, and cellular heterogeneity.
Integrated AI
Combine models, biological information, retrieval, and interfaces into larger biomedical AI workflows.
Biomedical Research Agent
Explore how an AI system can break a research question into steps, retrieve information, and organize evidence.
Synthetic Biology Copilot
Explore how language models might help organize biological design questions while keeping scientific evidence and uncertainty visible.
Drug Discovery Sandbox
Explore how computational filtering and ranking can reduce a large candidate space into a smaller set for further investigation.
Precision Medicine Explorer
Investigate how biomarkers and patient-level variables can be organized and visualized to explore differences across simulated cases.
The notebook isn't finished.
Labs is intentionally evolving. Some experiments are interactive now, others are being built, and some are still questions in my notebook.
The goal isn't to make clinical tools. It's to take complicated ideas in biomedical AI and turn them into things you can inspect, manipulate, question, and understand.