Challenge

SnapDragon Monitoring is a start-up that monitors the internet on behalf of brands, checking for any infringements of intellectual property rights. They offer this as a service to companies who wish to monitor online marketplaces for the sales of counterfeit goods. Snapdragon had attempted to build a machine learning tool to automate part of their service, in collaboration with a local university. The approach had worked in theory but had not been fully implemented and was not being used by the business. Ortom was approached to review the current approach and advise improving it and developing a strategy for effectively using machine learning to help scale the business.

Solution

Ortom conducted a full review of the business to understand exactly where machine learning could help and what had happened in the previous attempt. We conducted in-depth interviews with all of the people who were involved in the project and reviewed all existing code and documentation.

Result

We found that while some good work had been done, the team that was due to use the solution did not have faith that it was giving the right answers. We found a reliably simple solution to this problem by altering the way the output from the model was interpreted. This reduced the number of false negatives immediately. We then helped Snapdragon to hire a new data scientist to take over development of the ML solution. We helped to select CVs and interview candidates and finally recruited a high quality employee. Ortom worked with the recruit to handover and ensure that their first few months were as productive as possible.

Conclusion

Snapdragon was able to develop an ML driven product by working with Ortom. All the previous work did not go to waste and they were able to hire and make the most of a great new employee.