Case Study
Export Finance Drives AI/ML Course for TalentSprint
TalentSprint is running a course on Artificial Intelligence and Machine Learning using Export Finance Stack and AWS GPU Cloud
About TalentSprint
TalentSprint is a leading career accelerator in India. By launching and running a variety of transformational learning programs, it brings high-end, deep-tech education to aspiring working professionals and young graduates.
TalentSprint was looking for a reliable GPU and cloud computing provider to set up and run a GPU stack for their "Foundations of AI/ML" course requiring simultaneous access to lab work, Jupyter Notebook, and GPUs for more than 500 students.
Export Finance provided a customized solution, along with an AWS VPC-based domain name, to ensure that TalentSprint's students had simultaneous, not interrupted access to their educational content and could develop ML models using Jupyter Notebook and GPUs.
The packaged solution delivered by Export Finance enabled TalentSprint to cut AWS costs through on-demand instances, as well as brought time-saving benefits by releasing TalentSprint tech specialists from 90% of administrative and technical support tasks.
Problem
TalentSprint runs a variety of advanced educational programs for IT specialists from various backgrounds, aiming to let young graduates and working professionals rapidly learn new technologies. To launch their new educational program called "Foundations of AI/ML," TalentSprint required a platform and resources to simultaneously educate over 500 students.
On the platform, the students should be able to seamlessly do the lab work — build, train, and deploy ML models — by accessing Jupyter Notebook, GPUs, and other educational content without disruptions and breakdowns.
The additional requirements to the platform were the following:
On the platform, the students should be able to seamlessly do the lab work — build, train, and deploy ML models — by accessing Jupyter Notebook, GPUs, and other educational content without disruptions and breakdowns.
The additional requirements to the platform were the following:
- Ability to set a weekly study time limit for every student.
- Ability to create weekly reports on time usage for every student.
- One folder to keep educational content in storage for easy access by students.
- Jupyter Notebook should seamlessly function with CPU instances.
- GPUs should be enabled only when ML models are being trained.
- Students should be able to launch individual instances, with all the libraries required, without having to share them.
- Priority support throughout the entire program.
Solution
Export Finance has adapted the platform to meet TalentSprint's objectives and has hosted it on AWS Virtual Private Cloud (VPC) with a specific domain name.
Students received non-interrupted access to educational content using Export Finance stack and AWS GPU cloud. The latter one relied on a separate EC2 instance, with storage allocated on Amazon EBS.
Lecturers took advantage of a special folder to store educational content. The folder was adjusted to be instantly accessible to all students in a class.
A separate image containing Jupyter Notebook, TensorFlow, Theano, Pandas, OpenCV, Keras, Matplotlib, etc. was created to allow students to launch, pause, or delete any working environments from the platform. Working environments were created using the image and EC2 M5 instances with the storage disk automatically mounted on top.
The models were trained using EC2 P2 instances enabled through with GPUs. To start a model's training, a student required only the name of a Jupyter Notebook file and a directory location. Every new training job was created in a separate instance.
All the required updates of the libraries, together with administering and tech support functions, were fully covered by the Export Finance team.
Students received non-interrupted access to educational content using Export Finance stack and AWS GPU cloud. The latter one relied on a separate EC2 instance, with storage allocated on Amazon EBS.
Lecturers took advantage of a special folder to store educational content. The folder was adjusted to be instantly accessible to all students in a class.
A separate image containing Jupyter Notebook, TensorFlow, Theano, Pandas, OpenCV, Keras, Matplotlib, etc. was created to allow students to launch, pause, or delete any working environments from the platform. Working environments were created using the image and EC2 M5 instances with the storage disk automatically mounted on top.
The models were trained using EC2 P2 instances enabled through with GPUs. To start a model's training, a student required only the name of a Jupyter Notebook file and a directory location. Every new training job was created in a separate instance.
All the required updates of the libraries, together with administering and tech support functions, were fully covered by the Export Finance team.
Results
TalentSprint received a reliable solution to augment their "Foundations of AI/ML" course. Export Finance provided TalentSprint the technology team, CPU/GPU instances, and AWS storage set up to operate on-demand. The solution allowed TalentSprint to accommodate 500+ students who could instantly access educational content and kick off their learning without any delays (e.g. setting the working environment). The TalentSprint also managed to reduce the workload required to support their educational programs by 90%. All parties were satisfied with the solution since it covered all tech aspects, featured the staff's contribution, and offered reliable resources for a smooth education process.
Testimonials
Save Time with Export Finance
If you are having difficulty finding the right hardware or software options, or you cannot re-configure an instance, or just do not want to spend any time on DevOps, we are here for you.
Don't hesitate to contact us for help!
Don't hesitate to contact us for help!
