When the founders of cnvrg.io set out to develop its platform for data scientists, they aimed to make it useful for any industry and use case. They also wanted to make sure it was extremely accessible not only to data scientists and data learning engineers, but also software developers and analysts. They wanted to make it simpler for any enterprise to implement AI into their business whether they had a full time data scientist or not. cnvrg.io was excited to learn that it was a perfect solution for Israel Tech Challenge (ITC).
ITC is a training program that offers students advanced and intensive tech training focused on the most in-demand skills in tech such as data science, cyber security and software development. Through their elite data science fellows program students learn and practice all the in-depth skills necessary to be a data scientist in the industry. In fact, near the end of their training, they are required to prepare their own projects and build models on their own. The program all leads up to an internship where they can apply what they’ve learned.
Focusing on teaching tech with on-demand compute
ITC’s teaching method is very hands on, and designed on the belief that practice makes perfect. Though, with machine learning the practice aspect can be logistically complicated. Because there is a lot of experimentation related to machine learning and deep learning, a student cannot always work on their own computers or machines. The experiments that they needed to run were too complex to run day-to-day on student’s computers. Initially, they sought external help through cloud providers to assist them.
What they found was that it was not so trivial to use these external services without having a system to manage and communicate between the users and cloud providers. The teachers were burdened by having to manually fire up and down compute instances each time a student needed to run an experiment. And, for a class of 60 students, ITC understood they needed a better solution.
cnvrg.io provided the communication they needed between the cloud providers and students. Students could instantly fire up on their own compute, they could manage their budget and track how much they had spent on their work. Better yet, the spot instances integration gave students an extra 80% savings.
Supervising and empowering students through model tracking and management
Though unintended, cnvrg.io was excited to discover that their data science platform offered the perfect solution for ITC’s program. Not only did cnvrg.io automate the process of running experiments and setting up environments like Jupyter, but it also allowed teachers to manage the data science projects and keep track of all the students experiments. cnvrg.io tracked every stage and allowed teachers to identify student mistakes and help them fix it with real hands on learning. This helped support ITC’s hands on approach, without helicopter teaching.
Centralized hub for all machine learning datasets
Before cnvrg.io, class time was wasted waiting for students to upload large datasets individually. The process was frustrating, and was very vulnerable to tech related issues. With cnvrg.io, teachers could upload any size dataset to the platform just once, and the students would have it available to them and seamlessly use it any time they needed. Additionally, there was a lot of time saved from the cnvrg.io support team that was attentive and responded immediately to any inquiry.
We sat down with Ofir Chakon, ITC’s tech mentor who was responsible for the operation of the course, ensuring all systems were set to teach data science. We asked him to share his experience with cnvrg.io in the classroom, since it was something we hadn’t encountered with other customers.
“With cnvrg.io, data science students can fire up a machine any time they need to run an experiment, they can manage their own budget and determine what machines to use for each project. It’s a very convenient way for them to practice training and use machines in a seamless way, which is better than the manually intensive and expensive process we had before.” Ofir explained.
Using cnvrg.io to enhance data science learning
We asked Ofir what his advice would be to other data science programs considering integrating cnvrg.io, to which he said:
“I understand that ITC is not the standard user of cnvrg.io, as we are not a company or corporate data science team. But, in terms of other learning programs, I can say that we are grateful for the time it saved us. Especially because we know how much time it took in courses before cnvrg.io.”
While cnvrg.io is still built to help corporate data science teams, the cnvrg team has been excited to develop ITC’s partnership. ITC students have provided many suggestions for features which has really helped to expand the mission to make data science more accessible. cnvrg.io is now an integral part of ITC’s data science fellow program, making regular visits to help students integrate the platform for their specific needs.
Now, after two successful courses, it’s exciting to see the projects built by students and to follow them through their careers. Stay tuned for a report of demo day where students will present what they’ve accomplished using cnvrg.io. As it turns out, cnvrg.io is a great platform for data science courses too.