New Sense App

AI sound-therapy prototype, frequency tuner + shared listening sessions.

Role: Frontend
Year: 2024
Category: Mobile
Stack: React NativeRedux ToolkitJavaScriptiOSAndroid

02 · Outcomes

What the build actually shipped.

Numbers, not adjectives. The stats below are pulled from production. Anything still live, still measured.

Built

to spec

2024

Frontend

scope

mobile

tech in stack

React Native · Web Audio

2024

shipped

year of build

03 · Overview

What it does.

Pick the part of the body that hurts, and the app generates a frequency profile claimed to help. Whether the science fully holds up isn't the design problem, making the experience feel calm rather than clinical was. Bonus mechanic: invite someone else to listen to the same session in sync, so meditation becomes a shared thing instead of solo headphones.

05 · Process

How the build actually ran.

Four phases. Each one with a deliverable the client could see, not a slide deck of intent.

01
Conversation
Mapped the audience, the core flows, the constraints. Stack picks came after the user story — they followed what the product had to do, not the other way around.
02
Plan
Schema-first. Data flow, service boundaries, what gets cached and where. The boring decisions made on purpose so the build above them stays honest.
03
Build
Frontend, backend, deployment, the operational pages. Honest weekly demos, no Tuesday-before-launch heroics, no hidden scope.
04
Ship & Maintain
Store / web release cycles, monitoring, analytics, the dull work that decides whether a launch survives 48 hours. Documented handover.

06 · Decisions

The choices that shaped the build.

Tech is downstream of decisions. These are the calls made early enough to compound the right way. The ones a code review six months later still agrees with.

End-to-end ownership

One engineer from schema to deploy. Removes the most expensive failure mode in product builds — the seam between specs and code where intent dies.

Schema before UI

Data shapes first, screens second. The frontend is a view of the data, not a parallel reality. Saves weeks of rework when scope shifts.

Boring tech where it matters

Relational stores for payments and ledgers (ACID is non-negotiable), document stores for content, Redis for cache. Pick the dull tool for the dull job.