Local Tradesman Search

Objective

The primary challenge was to provide a fast and accurate search engine that can understand human speech due to the complexity of search queries the users asked. This way, we decided to develop a search engine powered by natural language processing and machine learning.

BENEFITS

We decided to develop the core of the application with Ruby on Rails, as due to the development speed that framework provides we were able to deliver the Minimal Value Product in the shortest terms. As the framework offers a variety of pre-built modules (so-called GEMS), it was easy for us to support and enrich the application over time.

The core of the ML-powered search engine was built with Python and TensorFlow.

Solution

Azati was hired to design and develop the software, both web and mobile frontend (iOS and Android) and server backend (administrative panel). Azati covered the entire development process, including project management, requirements gathering, specification writing, coding, testing and infrastructure setup. A few key competitive advantage ingredients:

Social networking model: it is a true community, where tradesmen maintain their customizable profile pages, describing their business, what they have to offer and their success stories from the clients, including completed job ratings.

Reputation based rating: reputations are built by the work that tradesmen do, not how much money they pay.

Loyalty program support: tradesmen get points for early registration upon the initial service launch, points for referrals and other campaigns, and these points can be used to cover the subscription fees.

The machine learning approach helped the user to find a local tradesman about 272% faster, in comparison to traditional search engine algorithms, what improved user satisfaction and positively affected search engine optimization.