Skip to content

Real infrastructure.
Open to everyone.

ndexr is live high-performance computing we run our own production work on. Sign in and the same platform is yours — yourname.ndexr.io , an agent, your code, and a live site. Describe what you want; it gets built on the real thing. Every version saved and reproducible.

Still in early access. Free to start, no credit card.

Most ideas never become real things. We change that.

You don't need to know how to code. You don't need a developer. Sign in, describe what you want, and it gets built — a real working site at your own address, yours to keep. Every version is saved and reproducible: you can always see how it was made, and rebuild it exactly. That's the same guarantee a pharma lab runs on — here it's on by default, for everyone.

Not a sandbox. Real infrastructure.

We build on this every day. So can you.

ndexr is real high-performance computing infrastructure — the same platform we use to run our own production work. Every site you spin up, every build, every deployment runs on the live system, not a demo of it. It's a working company you can actually log into and use today, not a waitlist or a pitch deck. The work we do to make our own infrastructure faster and more reliable is the same work that makes yours better — we're on the record building it in the open.

Reproducible by default

Built once. Verifiable everywhere.

Every package is built with EasyBuild on a pinned toolchain, content-addressed, and published to a cryptographically signed CVMFS repository. Every machine that mounts it fetches the exact same bytes — checked against the signed catalog on every read. The build you validate is the build that runs: on any node, anywhere, indefinitely.

R 4.6.0 Python 3.12 the scientific computing toolchain underneath Built to scale past 40,000 packages — each one reproducible
Pinned & content-addressed

Exact versions, exact toolchain, hashed. Nothing floats; nothing rebuilds differently tomorrow.

Signed & verified

Every catalog is cryptographically signed. Clients reject anything that doesn't match the trust anchor.

Identical on every node

No per-node installs, no rsync, no drift. The same tree mounts on a laptop and on a 1000-core cluster.

Use the stack

Mount it. Load it. Run it.

You don't install anything. Point the CVMFS client at our signed repository and the whole stack appears under /cvmfs/software.ndexr.io , fetched on demand and cached locally.

                  # 1 · install the CVMFS client
wget https://cvmrepo.s3.cern.ch/cvmrepo/apt/cvmfs-release-latest_all.deb
sudo dpkg -i cvmfs-release-latest_all.deb && sudo apt-get update
sudo apt-get install -y cvmfs && sudo cvmfs_config setup

# 2 · trust ndexr's signed repo
sudo mkdir -p /etc/cvmfs/keys/ndexr.io
sudo curl -fsSL https://cvmfs.ndexr.io/keys/software.ndexr.io.pub \
  -o /etc/cvmfs/keys/ndexr.io/software.ndexr.io.pub

# 3 · point the client at our Stratum-0
echo 'CVMFS_SERVER_URL="http://cvmfs.ndexr.io/cvmfs/@fqrn@"' | sudo tee /etc/cvmfs/domain.d/ndexr.io.conf
echo 'CVMFS_KEYS_DIR=/etc/cvmfs/keys/ndexr.io'              | sudo tee -a /etc/cvmfs/domain.d/ndexr.io.conf
echo 'CVMFS_REPOSITORIES=software.ndexr.io' | sudo tee /etc/cvmfs/default.local
echo 'CVMFS_HTTP_PROXY=DIRECT'             | sudo tee -a /etc/cvmfs/default.local   # single client, no Squid

# 4 · mount on demand — the whole stack appears read-only
cvmfs_config probe software.ndexr.io
ls /cvmfs/software.ndexr.io

# 5 · load a reproducible module and run
#     (exact MODULEPATH + every available module: software.ndexr.io)
module load R/4.6.0-foss-2024a
R
                

For anyone with an idea

Personal projects

A habit tracker. A quiz for your group chat. A game for your kids. A tool for your weird hobby. Describe it and it gets built — no technical knowledge needed.

Work & small business

A booking page. A staff schedule. A client form. Describe how you work and it gets built that way — not the way a template decides you should.

Curious minds

Want to understand how software gets made? Build something real, watch decisions happen, change your mind halfway through. The best way to learn is by making things.

How it works

1
Describe your idea

Describe what you want in plain words. Half-formed is fine. It can be a game, a tool, a form, a tracker — anything. Ask questions, get answers, figure it out together.

2
Watch it get built

It builds right in your browser, step by step — and every step is written down. You steer, you change your mind, you keep going. The real thing, not a mockup, with a record of exactly how it was made.

3
You own it

Your site lives at yourname.ndexr.io. Your data, your code, and the full history of how it was built. Walk away anytime — you take everything with you, and it rebuilds exactly the same, anywhere.

Every domain is a workspace

One address. The whole toolchain.

Each domain carries everything you need to build it: an agent to describe what you want, a terminal, a VS Code session, a Git repo, and the live site itself. Move from one.ndexr.io to another.ndexr.io and you're working in seconds.

Domains nest. That's what makes consolidation possible — aggregating scattered work across many repos and turning institutional knowledge into navigable, inspectable sites.

yourname.ndexr.io
├─ agent · describe → build
├─ terminal · a real shell
├─ code · VS Code session
├─ git · immediate repo
├─ site · the live URL

How it scales

Many addresses. One system.

n domains, n modules, one app. Every address points at one server running a single Shiny app. Inside that app, each site is just a module. nginx sits between DNS and the app and routes every request by its address — so what gets rendered is contingent on where you came in.

One repository holds all of it. You add a module, not a codebase — and the same app serves every company at once. Scale past one app and each tenant gets its own shared instance.

n domains1 nginx1 appn modules

n domains
company-a.com
company-b.com
yourname.ndexr.io
… and on
nginx
routes by request
one app · n modules
shiny app
├─ module · company-a
├─ module · company-b
├─ module · …

One repo. Every company a tenant. Scale to n apps and each gets its own shared system.

Why it's a better deal

No seats. No tiers. No ceilings.

Many platforms charge by the seat and keep the features you need behind upgrade tiers. We take a simpler approach: the capabilities are on by default, what you build is yours, and you can take it with you whenever you like.

The usual model

  • Priced per seat — the bill climbs with every person you add
  • Access controls and private sharing locked behind upgrade tiers
  • Custom domains rationed by plan: one, then five, then ten
  • Fixed RAM and CPU caps for every piece of content
  • Security, compliance and scale mean enterprise procurement and a sales call
  • Your work lives on their platform

With ndexr

  • No per-seat tax — share with your team, your customers, or the whole public
  • Access controls and private sharing included from the first day
  • Bring as many domains as you want, all under one system
  • Runs on real high-performance infrastructure, not a fixed tier
  • Validated, reproducible and on-prem capable — without a long procurement process
  • Your code, your data, your repo — export everything and walk anytime

For teams that need it inside their own walls

The whole platform, installed where you control it.

The reproducible, on-the-record builds everyone gets here are the same ones a regulated lab can validate. Everything that runs ndexr.io can run inside your own infrastructure — on-prem, air-gapped, or in your private cloud: many users and many applications under one validated install, where every version has to be accounted for.

Reproducible builds

Every dependency pinned, every image content-addressed. The build you validate is the build that runs — bit for bit, every time, on any node.

Many users, many apps

One platform serving isolated tenants. Each user gets their own apps, data schema, and access boundary. Scale horizontally across an HPC cluster.

Validation-ready

Designed for GxP and other regulated environments. Audit trails, deterministic deployments, and documented build provenance to support IQ/OQ/PQ.

The same platform we run in production — installed and validated inside yours.

Talk to us about a private install

Get started

What do you want to make?

Still working through a few things, but the door's open. Come in and try it.

Get started free

Free to start. No credit card.