While yet being part of deepsense.ai team, we have real fun challenging ourselves by taking part in global data science competitions, like Kaggle.
We realised that the more experiments we we able to run, the better machine learning results we achieved.
We’ve developed a tool for internal usage to be able to achieve such high results under the same time and budget constraints. Using our tool, which we then called Neptune, we were able to conduct over 1600 experiments, share infrastructure, collaborate everyday and track the entire process.
That enabled us to build a high-quality machine learning model and win Kaggle’s Right Whale Recognition challenge.
Due to the efficient infrastructure utilisation and cut of IT administration work we were able to save about 15% of our data scientist team time. Since then we have been using Neptune to all our projects.
In November 2017 we decided we want to focus on developing the tools which will push AI forward. Today neptune.ml it’s not only a platform, it’s a place for everyone who want to develop in data science, by using efficient processes, collaborate and share knowledge with community. Neptune.ml it’s a way to do data science smarter, not harder.
Originally published at neptune.ml.