Skip to content
Building frontier models for biology

The next generation of single-cell models.

F15 Labs is building the next generation of single-cell frontier models in biology — foundation models that learn from cellular atlases to understand cell states, predict perturbations, and accelerate discovery.

Models

Frontier models for single-cell biology

We're training the models that will define how biology is read at cellular resolution — from raw expression profiles to actionable predictions.

Foundation models

Pretrained on large-scale single-cell atlases — scRNA-seq, spatial, and multiome — to capture gene regulation, cell identity, and tissue context.

Cell state understanding

Rich embeddings for annotation, trajectory inference, and cross-dataset integration — turning millions of cells into interpretable biological insight.

Perturbation prediction

Simulate drug, genetic, and environmental perturbations at single-cell resolution — predict how treatments reshape cell populations before the experiment.

Approach

From atlases to insight

01

Curate

Aggregate and harmonize single-cell datasets across tissues, species, and modalities — building the atlases our models learn from.

02

Train

Scale foundation model pretraining on billions of cells — learning gene programs, cell-type hierarchies, and perturbation responses.

03

Discover

Deploy models for embedding, annotation, and perturbation prediction — giving researchers tools to move from data to hypothesis faster.

Stack

Built on the best of single-cell and AI

We combine state-of-the-art single-cell methods with frontier model architectures — pretrained on atlases, fine-tuned for your biology.

TransformersGraph neural networksVariational autoencodersDiffusion modelsContrastive learningSelf-supervised pretrainingEncoder-decoderAttention mechanismsGenerative modelsMulti-modal fusionMixture of expertsFine-tuningTransformersGraph neural networksVariational autoencodersDiffusion modelsContrastive learningSelf-supervised pretrainingEncoder-decoderAttention mechanismsGenerative modelsMulti-modal fusionMixture of expertsFine-tuning

Expression & sequence

Gene-level transformers and variational autoencoders trained on transcriptomic profiles across tissues and conditions.

Scale & pretraining

Distributed training on curated single-cell atlases — harmonized metadata, quality control, and cross-dataset integration at billion-cell scale.

Inference & discovery

Embedding APIs, perturbation simulators, and evaluation pipelines — so researchers get answers, not raw matrices.

Team

The people building F15 Labs

A small team at the intersection of single-cell biology, foundation models, and software engineering.

Arthur Tao

Arthur Tao

Co-founder

Addison Crider

Addison Crider

Co-founder

Glenn Shields

Glenn Shields

Co-founder

Contact

Talk to us

We're partnering with research groups and biotech teams pushing the frontier of single-cell biology. If that sounds like you, say hello.