KOGU
Kogu organizes ML experiments by logging your code, parameters, data, and results. It's fast, and requires minimal overhead

WHY KOGU
Transparent

experiment
management
See the big picture, including details.
See all your experiments and their performances. Select the optimal one to be deployed in
your product.
Easy

reproducibility
Remember used data,code,and parameters.
Share knowledge and reproduce experiments with the same data, algorithms,
and parameters
Clear

visualization
See the status and performance of your model
Visualize model's performance and stop the training when
overfiting.
KOGU FEATURES
Efficiency
Focus on execution and delivery, not on the past. Kogu logs the environment, data, script, parameters, and outputs for you to easier find, compare and reproduce experiments.

Transparency
For data scientists and for managers. Kogu is your single point of truth where you can compare experiments and share them with your team.

Knowledge base
Unified project structure combined with loggings will not replace you, however, will make it easier to make sense out of your work when you are gone

Clear visualization
Need help understanding the status of your trained model? Kogu helps you visualize the parameters you need. Just select them and we do the rest.

SUPPORTING ALL FAVORITE LIBRARIES





TensorFlow
Keras
Sckit-learn
Pytorch
Torch
ABOUT KOGU
Kogu started as an internal project for Proekspert's data science team. We learned it the hard way; having several projects, training different models, optimizing hyper-parameters, deploying models and improving them is a repeatable activity. Losing only one of the data science pillars (data, script, parameters, outputs) reflects in wasting your time in searching and re-doing experiments almost from scratch.
Small steps in organizing data science workflow can have a substantial impact on your productivity, in our case approx. 25% of the time. There are several projects tackling different features. However, our mission is to make those features stable and seamless to use, so data scientists can focus on what matters the most,
building complete AI solutions.