Case Studies

Dataffirm

Frontend development for Dataffirm's dataintensive Angular 2 investment platform, translating machinelearning outputs and live market, company, news, and sentiment data into responsive dashboards, search tools, and interactive visualisations.

Screenshot of the Dataffirm website; part of John Kavanagh's selected project work.

In Brief

Dataffirm was a fintech startup building an alternative investment platform powered by machine learning, big data, live market activity, social signals, and local news sources. The product helped investors search, compare, and analyse companies and people through realtime dashboards, data visualisations, and investment signals.

My Role

I joined Dataffirm during the early postfunding stage as one of two frontend developers. Working closely with designers, analysts, and a globally distributed backend and data science team, I helped build a responsive Angular 2 application that turned machinelearning outputs, big data, live market signals, online sentiment, and company profile data into usable product interfaces across multiple iterations.

Technologies

dataffirm.com
  1. HTML5 & SCSS
  2. Angular
  3. D3
  4. Node.js
Graphical representation of a network in multiple colours on a purple background.

In Detail

The product began as an evolving MVP: a responsive Angular 2 platform using Sass, D3, and Node.js, built around realtime market data, machinelearning outputs, and large structured datasets. The challenge was to turn those signals into interfaces investors could search, compare, and act on.

This was machinelearning and bigdata product work before "AI" became the default label for every dataled feature. The value was not the terminology; it was the attempt to surface useful investment signals from live market activity, company data, media mentions, location data, and online sentiment.

Because the product direction was still developing, the application moved through several iterations shaped by investors, stakeholders, and user feedback. The front end therefore had to remain flexible enough to absorb substantial change without making future expansion harder.

The application combined complex data tables, dashboard panels, search tools, and interactive visualisations. It was primarily designed for desktop use, where users could absorb more information at once, whilst mobile and tablet access still needed to remain usable in a reduced form.

The search functionality within the platform was separated into several facets, ranging from a straightforward textbased search for companies or officers up to much more complex dimensions involving location, media mentions, or even just how active they were on Twitter...

Screenshot of the Company search results table, desktop screen size.Screenshot of the Company search results page in map view, desktop screen size.

Company and Person Profiles

Using masses of data, the platform presents Company and Person profiles in a datadriven dashboard style, which quickly allows the visitor to get an overview of their position, activity in the media, elsewhere online, and the opportunity to investigate further. This panelbased dashboard is customisable by the user; they can determine which data falls where on the page.

Screenshot of an individual's profile displaying mentions and features in the press and social media, desktop screen size.Screenshot of a Company's profile page displaying location map and key financials. Desktop screen size.

Content Pages

Simple, accessible, contentmanageable pages were developed in various layouts and with a small component library. In this example, a Knowledge Centre was put together to offer advice and guidance to platform users, including text, imagery, and video.

Screenshot of the Help Centre search results page, tablet screen size.Screenshot of the Help Centre search results page, mobile screen size.

Have a similar
platform challenge?

I help teams make difficult React, Next.js, migration, performance, and technical SEO work clearer, safer, and easier to deliver.