Building Dhwani: How DeployByDesign Supports an AI-Powered Digital Stethoscope

Building Dhwani: How DeployByDesign Supports an AI-Powered Digital Stethoscope

---

Diagnosing heart murmurs isn't always straightforward. Some murmurs are harmless, linked to things like anaemia or hyperthyroidism, while others point to more serious conditions such as valve stenosis, regurgitation, or congenital heart defects. Telling the two apart quickly and accurately is exactly the problem Dhwani is trying to solve — an AI-powered digital stethoscope designed to help doctors catch these differences earlier and with more confidence.

Building something like this isn't just a software problem. It requires training models on acoustic data, refining them carefully, and making sure the results are precise enough for doctors to actually rely on. That kind of work needs solid infrastructure behind it — which is where DeployByDesign came in.

---

What the Project Needed

Getting Dhwani's models to a reliable standard meant the team needed:

• GPU compute for training and testing audio-based diagnostic models
• A workspace that made it easy to experiment and iterate quickly
• Secure handling of data, given how closely the project touches patient diagnostics
• A setup that didn't require a dedicated infrastructure team just to keep things running

---

How DeployByDesign Helped

GPU Access Without the Hassle

Training models on acoustic data isn't light computational work. DeployByDesign gave the team GPU compute whenever it was needed, without the usual overhead of provisioning servers or managing infrastructure — so the focus stayed on improving the models themselves.

A Simple Environment for Fast Iteration

A lot of the work on Dhwani involves trying things, checking results, and adjusting quickly. DeployByDesign's ready-to-use workspace meant the team could get started on a new experiment in minutes instead of losing time to setup.

Built-In Security for Sensitive Work

Since Dhwani deals with data close to patient health, having governance and access controls in place from the start mattered. DeployByDesign made that part of the foundation rather than something to figure out later.

Less Time Spent on Infrastructure, More Time Spent on the Model

With the compute and environment side handled, the team could spend more of its time actually working on accuracy — which, for a project like this, is the part that matters most.

---

Wrapping Up

Dhwani is still a work in progress, but it's a good example of what happens when infrastructure doesn't get in the way. Having GPU compute available on demand, a workspace that's ready when the team needs it, and security built in from the start let the team spend its energy where it counts — on building a tool that can genuinely help doctors make better, faster diagnoses.