Front end · React · TypeScript · AI-assisted
Telematics Front End
I was hired as the lead front-end engineer to turn a set of prototypes into a coherent React and TypeScript codebase for a SaaS platform that US fleet operators use to see their telematics data. I deleted what was dead, gave the rest one structure, and built the test suite from scratch.
What it does
The front end of a fleet analytics product, rebuilt so a team could change it without fear.
Fleet data in the browser
US fleet operators open the app to see their vehicles' telematics data. The analytics are Sisense dashboards, rendered as native React pages through the Sisense SDK.
One codebase, one way of doing things
The prototypes came from an outside team who knew the API but not React. Now every feature follows the same structure and reuses the same shared components.
Tests that make change safe
There were no front-end tests at all. I built the unit test suite from scratch to 98% coverage, and regressions became far less common.
Before and after
Separate prototypes became one app: feature folders on a shared component library, app state in React Context, and data from the Sisense SDK plus a few ASP.NET endpoints.
Technology stack
React and TypeScript on the front, the Sisense SDK for analytics, a thin ASP.NET layer for the rest, and AI tools for the second half of the build.
Engineering highlights for technical reviewers
The decisions that turned prototypes into a product.
Delete first
The fastest way to make the code clearer was to remove what wasn't used. I took out a large amount of dead code and files before restructuring anything, so the reorganization only had to deal with code that mattered.
Features plus shared components
Each product area lives in its own feature folder. Anything used in more than one place moved into a shared component library, so the same control looks and behaves the same everywhere.
React Context, not Redux
The Sisense SDK already caches query results, so the app didn't need a heavy state library. Shared app state lives in React Context: less code, less to learn, nothing extra to keep in sync.
Sisense as the main data layer
Most pages query Sisense directly through its SDK and render the results as React. Where the app needed its own data, I added calls to ASP.NET controllers backed by SQL Server.
A test suite from zero
No tests existed. I built the front-end unit test suite from scratch to 98% coverage, which made the rest of the cleanup safe and kept regressions down afterwards.
Hired to lead it
I joined as the lead front-end engineer specifically for this rebuild, and set the structure and patterns the front end is built on.
By hand, then with AI
The project spans my switch to AI-assisted development.
First 3 months: by hand
The cleanup, the new structure and the first patterns were written the traditional way, file by file.
Last 6 months: Cursor and Claude Code
The rest of the build was driven by Cursor and Claude Code. Today I work in Claude Code only.
Capabilities demonstrated
What leading this rebuild proves I can deliver.