With our promise of early, rapid delivery, our past clients over-delivered on their internal project metrics and timelines.
We built a model that helped the client predict how long a property would stay on the market at a given price. This enabled the client to purge their portfolio and acquisition pipeline, avoiding multi-year holding periods. Aside from the model, we also improved on divergent and incomplete data sources.
After the client acquired a competitor with millions of patients, we connected the disparate data sources together, de-duplicated customer records, and built both simple dashboards and more complex statistical products. This enabled the client to profit from synergies beyond the price paid for the acquisition.
The client wanted to automatically classify product images, but none of the commercially available machine vision models had been trained on the particular products sold. We built and integrated a custom machine vision algorithm, including the facilities for the staff to train examples in order to continuously improve the algorithm.
The product was used daily by 4 million customers and new product features were several months behind schedule due to scaling up of client’s requests. We helped unblock the launch of the next product version, and implemented features such as natural language processing and analytics.