To achieve the highest performance for your dataset, there is no substitute for a custom neural network which is designed and trained for your specific application and dataset.
"Off-the-shelf" pre-trained models can perform very well for select categories of applications. But, we have found that enterprise applications rarely fall completely into those categories.
We leverage publicly available knowledge when it provides a good starting point. We're happy to discuss the pros and cons of pre-trained networks with you to help you make your decision.
Have a team in house, but need more experts?
We’re not shy about teaching. We believe that the industry is still learning to crawl and we can all benefit and learn from each other.
When we engage with your team, every effort will be made to include them in the development process. We believe in transparency and deep collaboration.
After we deliver a neural network, we offer support and maintenance agreements and make ourselves available for questions from your team.
Custom because every project is unique.
We don't have a standard engagement model. We tailor the staffing needs to each project whether you just need a few hours of meetings, or a full featured production neural network.
After we understand the task at hand, we'll create an engineering resource plan and schedule. We will identify the higher risk portions of the plan and explain strategies to mitigate the risks. With your feedback, we'll revise the plan until you are satisfied with the costs and level of risk.