# UMAP Projects at AI Tinkerers

> Canonical HTML: https://aitinkerers.org/technologies/umap
> Markdown URL: https://aitinkerers.org/technologies/umap.md
> Technology record last updated: 2026-09-18T15:13:52Z
> Generated: 2026-09-22T00:32:26Z

UMAP (Uniform Manifold Approximation and Projection) is a non-linear dimensionality reduction algorithm that efficiently projects high-dimensional data to 2D or 3D, preserving both local and global data structure.

UMAP is a high-performance dimensionality reduction technique, grounded in manifold learning and topological data analysis. The algorithm constructs a high-dimensional graph (a fuzzy simplicial complex) representing the data's structure, then optimizes a low-dimensional projection to maintain that structure as closely as possible. Critically, UMAP offers significant advantages over older methods like t-SNE: it is substantially faster and more scalable, projecting the 70,000-point MNIST dataset in under 3 minutes, and excels at preserving the global, not just local, topology of the data. This makes it the gold standard for visualizing complex embeddings, identifying clusters, and serving as a robust preprocessing step in machine learning pipelines, particularly in fields like single-cell biology.

- Official technology site: https://umap-learn.readthedocs.io/en/latest/
- Public AI Tinkerers demos and talks: 5
- Result page: 1 of 1

## Recent Public Talks and Demos

### [MOFUX: vibe coding scientific data visualization](https://atlanta.aitinkerers.org/talks/rsvp_vRwTk9Be8_I)

This project is an interactive visual interface into a dataset of 220k experimental and generated Metal-Organic Frameworks (MOFs). The aim of the project is to give the materials science community an intuitive way to explore this open-source dataset.

- Event context: AI Tinkerers Atlanta: Code-First Demos & Technical Deep Dives | January 27, 2026 — 2026-01-27 — Atlanta
- Public talk page: https://atlanta.aitinkerers.org/talks/rsvp_vRwTk9Be8_I

### [Datamaps are 2D maps which visualize hyperdimensional latent spaces](https://seattle.aitinkerers.org/talks/rsvp_K_XKHiLogDE)

[I was talking to Joe at the end of the recent gathering at Fenwick & West on the 29th, "AI Trust & Security in AI: Demo Night." He said I should submit a datamap demo proposal. So, here it is.] Datamaps are a new visualization technique custom designed for vector databases, RAG systems, or more generally any collection of embedding vectors, that is any "AI-ready data." The best-of-breed datamap tooling is all permissively licensed open source such as UMAP, HDBSCAN, Tomonymy, DataMapPlot, etc. This lightning talk explains what datamaps are by providing a live interactive demo of Ai2 Semantic Scholar's curated library of scientific papers. The demo shows to how to DIY code up datamaps via open source. The (pre-cooked) example datamap would be a single static HTML page which each attendee can simultaneous load onto their phone and interact with. Here is the datamap they would experience: https://connoiter.com/datamap/55948af4-903c-41cb-997e-a1dfda45eb5f/

- Event context: Frontier Builds Demo Night: Experiments at the Edge of AI — 2025-11-13 — Seattle
- Public talk page: https://seattle.aitinkerers.org/talks/rsvp_K_XKHiLogDE

### [A Google-Like Search Experience for Your Photo Library with Opensource Tools](https://raleigh.aitinkerers.org/talks/rsvp_EEHGIBjFDGQ)

This demo showcases a photo album curation workflow using FiftyOne’s Brain similarity indexing and multimodal search. Images are embedded and indexed once via compute_similarity(), enabling rapid retrieval of visually similar photos and natural language search (e.g., “sunset at the beach”) using CLIP, which aligns text and images in the same embedding space. Users can sort albums by similarity to a reference image or a text prompt, find duplicates, and surface on-theme groups for curation. Low‑dimensional embedding visualizations further reveal clusters and outliers that guide pruning and organization. The experience is fully interactive in the FiftyOne App, and scalable via vector database integrations when working with large libraries.

- Event context: AI Tinkerers - Raleigh Inaugural Meetup (September 2025) — 2025-09-30 — Raleigh
- Public talk page: https://raleigh.aitinkerers.org/talks/rsvp_EEHGIBjFDGQ

### [Google Maps for Vector Embeddings](https://boston.aitinkerers.org/talks/rsvp_qMsq4QJj4uI)

Vector embeddings are powerful but nebulous. They are essentially high-dimensional arrays of floating point numbers that encode semantic meaning. But vector embeddings are hard to visualize particularly in context of other embeddings. But we can transform vector embeddings into 2D space and then graph them. By using UMAP we can preserve the relationship between high-dimensional vectors in 2D space. UMAP constructs a high-dimensional graph of the data, then optimizes a low-dimensional representation to preserve both: Local structure: Keeps nearby points together Global structure: Maintains relative cluster positions Non-linear: Can capture complex manifold structures Topological: Based on manifold learning theory

- Event context: AI Tinkerers Boston Meetup at Publicis Sapient — 2025-09-29 — Boston
- Public talk page: https://boston.aitinkerers.org/talks/rsvp_qMsq4QJj4uI

### [Contrasting Language Omics Pretraining](https://lausanne.aitinkerers.org/talks/rsvp_ZjMIQM_LY9M)

CLOP is an adaptation of OpenAI's CLIP but for Omics - in this demo, genomics. The model is trained on fasta, bed and gff files (representing genomes of different species and their annotations) to learn meaningful representations for further retrieval, classification and generation purposes. The model embeds DNA sequences according to species and biotype (e.g. exon, long non coding RNA, pseudogene, etc.)

- Event context: AI Tinkerers Lausanne June 2025 Meetup — 2025-06-16 — Lausanne
- Public talk page: https://lausanne.aitinkerers.org/talks/rsvp_ZjMIQM_LY9M

## Related Technologies

- [CLIP](https://aitinkerers.org/technologies/clip) ([Markdown](https://aitinkerers.org/technologies/clip.md)) — 10 public demos
- [Cohere embed-v4](https://aitinkerers.org/technologies/cohere-embed-v4) ([Markdown](https://aitinkerers.org/technologies/cohere-embed-v4.md)) — 1 public demo
- [D3](https://aitinkerers.org/technologies/d3) ([Markdown](https://aitinkerers.org/technologies/d3.md)) — 6 public demos
- [DataMapPlot](https://aitinkerers.org/technologies/datamapplot) ([Markdown](https://aitinkerers.org/technologies/datamapplot.md)) — 1 public demo
- [DDBJ](https://aitinkerers.org/technologies/ddbj) ([Markdown](https://aitinkerers.org/technologies/ddbj.md)) — 1 public demo
- [ENA](https://aitinkerers.org/technologies/ena) ([Markdown](https://aitinkerers.org/technologies/ena.md)) — 1 public demo
- [Ensembl](https://aitinkerers.org/technologies/ensembl) ([Markdown](https://aitinkerers.org/technologies/ensembl.md)) — 1 public demo
- [FiftyOne](https://aitinkerers.org/technologies/fiftyone) ([Markdown](https://aitinkerers.org/technologies/fiftyone.md)) — 2 public demos
- [GenBank](https://aitinkerers.org/technologies/genbank) ([Markdown](https://aitinkerers.org/technologies/genbank.md)) — 1 public demo
- [Github Pages](https://aitinkerers.org/technologies/github-pages) ([Markdown](https://aitinkerers.org/technologies/github-pages.md)) — 3 public demos
- [HDBSCAN](https://aitinkerers.org/technologies/hdbscan) ([Markdown](https://aitinkerers.org/technologies/hdbscan.md)) — 1 public demo
- [IGV](https://aitinkerers.org/technologies/igv) ([Markdown](https://aitinkerers.org/technologies/igv.md)) — 1 public demo
- [JBrowse](https://aitinkerers.org/technologies/jbrowse) ([Markdown](https://aitinkerers.org/technologies/jbrowse.md)) — 1 public demo
- [NGLViewer](https://aitinkerers.org/technologies/nglviewer) ([Markdown](https://aitinkerers.org/technologies/nglviewer.md)) — 1 public demo
- [ONNX](https://aitinkerers.org/technologies/onnx) ([Markdown](https://aitinkerers.org/technologies/onnx.md)) — 83 public demos
- [Pinecone](https://aitinkerers.org/technologies/pinecone) ([Markdown](https://aitinkerers.org/technologies/pinecone.md)) — 25 public demos
- [Python](https://aitinkerers.org/technologies/python) ([Markdown](https://aitinkerers.org/technologies/python.md)) — 662 public demos
- [PyTorch](https://aitinkerers.org/technologies/pytorch) ([Markdown](https://aitinkerers.org/technologies/pytorch.md)) — 273 public demos