Just a hobby project. Let’s talk.
Complex models. Reproducible science.
AI is changing how we discover and develop methods in biology. As researchers, we see practical gaps in hosting large, complex models and reproducing AI-driven analyses. This is a place to share what we’re learning and building.
In our experience, biological models don’t always fit a simple PyTorch setup. Custom data preparation, loaders, dependencies, and frequent updates can make them difficult to host—especially during development.
We’re exploring NVIDIA-native hosted GPU infrastructure on demand, to make complex models easier to build and run, with access to the compute we need, when we need it.
In agentic workflows, the same prompt can lead to different results. As AI takes on more analysis, understanding how a result was produced becomes essential.
We want to share ways to make these workflows reproducible: recording decisions and why they were made, preserving environments and versions, and creating notebooks that make results easier to rerun and check.