I am implementing an AI solution for one of MongoDB’s products and I faced the same problem that every engineer in my field faces - how do you get enough data to train your neural nets? Collimator proved to be the best solution for me because of how easy it was to get data from our back end, use that to create synthetic data, and then move that data back into my neural net. I’m now experimenting with training my neural nets within Collimator

I am using Collimator to build a proof of concept autonomous robot. I started with building the perception system using ML and after deciding that the Matlab license was too expensive for me, I decided I needed to develop in Python where I start with open-source models and build on them. I looked at Jupyter, Eclipse, Pycharm and AWS Cloud9 but constantly ran into issues because writing the perception systems took a bunch of time and training my models was too expensive and cumbersome. Instead of it being a pet project, it became a main project. I found Collimator while researching alternative tools and I’m glad I did. I’ve used Collimator to design my system. The interactivity between the Python and graphical environment is exactly what I needed. The error messages are easy to understand and debug. Plus, I no longer have to set up AWS clusters to run my model online.

I am a researcher in Vienna and have been using Matlab for many years. My lab decided to make the transition to Collimator a few months ago and I am very glad that we did. Collimator is by far the best modeling and simulation software that I have used for the following reasons: Getting help in Matlab and C was becoming more challenging as my project grew in complexity. Over the past few years, I felt like I was spending more time learning about the complexities of Matlab rather than doing my actual research. Using the block diagrams that Collimator provides has allowed me to simplify my code and shift the time I was spending writing convoluted scripts back to my research Since Covid, it has been difficult to work with some of my colleagues since we have to alternate days in the lab. This meant that getting feedback would take about 2-3 days until my colleagues came in. This slowed our research down quite significantly. However, now with Collimator I can continue to work together with my colleagues who are remote. I will usually tag them directly in my model for feedback and they get the notification instantly. It feels like we are in the same room