Why we built Tunnelr?

We were building a HIPAA-compliant CRM for doctors.

A big part of the system relied on GPU-heavy workloads. Running those on cloud GPU instances was expensive, so we used local GPU machines instead.

That solved cost and control.

But it created a new problem:
we needed to expose parts of this app to the outside world.

The usual tools (ngrok, Cloudflare Tunnel) worked technically,
but they introduced third-party infrastructure into a system handling sensitive data.

That was a non-starter.

We wanted:

• local GPU machines

• development on localhost

• external access when needed

• everything staying in-house

So we built a small internal tunneling tool that:

• runs on our own VPS

• uses our own domain

• forwards traffic directly to localhost

• supports path-based or subdomain routing

• has no timeouts or external dependencies

That tool eventually became Tunnelr.

We open-sourced it because this pattern keeps showing up:
local compute + strict compliance + real-world integrations.

Curious if others are solving this differently.