Salpa
The desktop app. Build a pipeline as a node graph, run it on your machine, and export it as one portable file.
salpa.app ↗BoundaryComputing
Open, visual workflows for scientific computing
We build Salpa — a desktop app that turns scientific computing pipelines into shareable, reproducible node graphs.
Salpa 0.3.0 · macOS · Linux · Windows · free
The idea
The workflow is the least-shared part of computational science.
The chain of tools behind a result is rarely reproducible, portable, or easy to reuse. Salpa makes it a visual node graph — legible at a glance, runnable by anyone, and reproducible by construction. Open workflow, alongside open data and open source.
Products
One platform, four pieces.
The desktop app, an open collection of workflows for it, managed GPUs for the heavy steps, and a CLI for building your own nodes.
Salpa Hub
Our official, open collection of node packages — real workflows, real licenses, installable from inside the app.
Explore Salpa Hub → Boltz-2 Chai-1 DiffDock ProteinMPNN Evo25 GPU nodes · live
Salpa Compute
The heavy steps run on our managed GPUs, from your local workflow. Free tier, no cloud account.
See Salpa Compute →
Salpa CLI
pip install salpa-cli. Scaffold a node package, test it in its own environment, and push it into the app.
See Salpa CLI →Salpa Hub
Explore Salpa Hub →Real workflows, ready to run.
Our own collection of scientific node packages — built, tested, and maintained by us. Install into Salpa and run.
metalparm-vwf
Take a metal complex to a QM-derived force field and an MD-ready topology — the EasyParm pipeline plus metallopeptide fragment fusion, as 20 nodes.
metaldock-vwf
Dock metal complexes with metal-aware QM charges — the MetalDock pipeline, as 6 nodes.
pdbmdauto
Automated protein preparation: a PDB in, an MD-ready GROMACS system out — 13 nodes.
Latest
All news →What just happened.
The key things, from the build log. Every small step in between is there too, in order.
Jacob Woltjer joins as cofounder
Jacob Woltjer joins BoundaryComputing as cofounder, leading business development. He comes to it as a user first: eight months building real workflows in Salpa for his MSc thesis — AI protein generative models, quantum-chemistry force-field parameterization and all-atom MD in a single pipeline — defended in July with a top mark.
Get involved →Salpa Hub is open source
Our official, open collection — the salpa-hub repository — is public. Install ready-made pipelines, or contribute your own.
See Salpa Hub →
Salpa at EuroBIC18
At the 18th European Biological Inorganic Chemistry Conference in Groningen, our founder Xinmeng Li presented Salpa — an open visual workflow platform that turns multi-step simulation pipelines (quantum chemistry, force-field generation, MD, docking) into interactive node graphs a researcher can run without scripting. The talk walked through two metal-chemistry workflows built entirely from open methods.
See the slides ↗Latest release · Salpa 0.3.0 · 2 Sep 2026 · in the log →
Vision
Three commitments, one direction.
BoundaryComputing builds open, reproducible tools for molecular science. We organise the work around the workflows we open, the infrastructure we build, and the complexity we decompose. Salpa grew out of real research pipelines — rebuilt so they can be seen, shared, and re-run. Early, open, and built with the community.
Try Salpa — free, on your machine.
Salpa 0.3.0 for macOS, Linux, and Windows. Compose and run scientific computing workflows locally, and export any of them as one file. No account, no lock-in.