Hewlett Packard Enterprise Data Science Institute

As AI and machine learning become a growing presence in society, the need for data science knowledge in major industries is expanding rapidly and globally. This year, the HPE DSI continues to support and strengthen the role of UH research in important global challenges by promoting community, industry, and institutional partnerships within its events, programs and resources.

At the forefront of the HPE DSI is Data Scientist Ishita Sharma. We asked her to give us some insight into the institute’s plans for the 2023-2024 year.

What do you do for the HPE DSI?

I work as a data scientist at the DSI, so I have many different hats. I provide consultation to researchers across the university in case they need any help in data science projects. I mentor students in a couple of different data science courses across the university, which includes the Sugarland campus and the other branches as well. I also take care of our workshops, helping to organize and host hybrid and web-based lectures which are related to data science, AI, and machine learning. We have been collaborating with different (UH) departments and industrial partners, such as, NVIDIA, HPC Society, Mark III Systems, HPE and MATLAB. I also do a little bit to help our users on our clusters.

What exciting events/programs does the institute have in store for the academic year?

I would say that we have wonderful upcoming workshops. We are trying to—for our Mark III AI/ML education series—collaborate with a student organization, Cougar AI. We aim to reach out to more on-campus communities for the purpose of engaging with data science.

What is the Micro-credential in Data Science Program, and how can students get involved?

We have two micro-credentials now: one is our Micro-credential in Data Science program and the other one is our Micro-credential in Advanced Data Science program. Those two programs are based on data science and machine learning. The first program, which is approximately three years old now, has been receiving a lot of applications—more than 30 per year. It’s a very popular program, so we offer full courses in Python. It starts with Scientific Programming with Basic Python, which is a very extensive course taught by Jerry Ebalunode, Ph.D. Then, I teach a course called the Principles of Data Management. It’s all about how you can get started with data science projects, from data pre-processing to transforming and building models. We also offer Introduction to Machine Learning, and that course is taught by Dr. Ioannis Konstantinidis. He covers lots of algorithms in machine learning. The last course is Data Analysis and Visualization Using ParaView and Tableau. All these courses are 15 hour-long courses. They are free for the UH community, but alumni or outside participants have to pay $250 per course.

What is the difference between the two micro-credentialing programs?

Micro-credential in Data Science serves as an introduction to data science and machine learning, whereas the second certification is more advanced. There, students learn deep learning and advanced data science. We also offer a course that teaches data analysis in R. Introduction to Machine Learning overlaps with the two certifications.

Tell us about your experience in teaching for the Micro-credential program.

I teach Principles of Data Management and I also co-teach Data Analysis and Visualization in Tableau and ParaView. If we’re talking about my experience in teaching, it’s been great. I’ve seen multiple students and faculty from different departments all over the university taking these courses. It’s welcome to anyone who wants to get started with data science. We try to teach skills rather than theories, so it can help (participants) clear their interviews and get started with their jobs or internships. These courses are not part of a degree program and won’t be displayed on transcripts, but we do provide digital badges, so that (employers) know that the students are certified in these skills. Graduates can put them on LinkedIn or whatever social profile they have.

The HPE DSI offers high-performance computing clusters in the Research Computing Data Core (RCDC). Could you explain these clusters, and how they can help in research for UH faculty and staff?

We have four clusters. The three main clusters, Opuntia, Sabine, and Carya, are used exclusively for research. They are all computationally effective clusters, so they boast both GPUs and CPUs. There are different numbers of processors, which are present across the clusters. Apart from that, we do have the last cluster, which is a training cluster. It’s for students and researchers across the university who want to get trained on our main clusters. One thing is that the main clusters are free for UH researchers only. The researchers can sponsor it to their students, but students cannot directly access them. Instead, they can access our training cluster.

What aspect of the supercomputing clusters do you think will benefit those using it the most?

If you have any research which is related to image analysis or image segmentation—the broader umbrella term would be deep learning research—you have to use GPUs, where there is an advanced computing need. There you can get started with the clusters, so that your research, whatever you are trying to create or train, can be trained much faster for free. You do not need to buy or set up clusters in your own lab; we have set it up for you at DSI so you can use GPUs that are available–although there is a cap, it’s available for free.

Sharma has also collaborated on several DSI projects for research, including maintaining a comprehensive database for the College of Medicine and creating an efficient data management system for the College of Education. Moving forward, the HPE DSI will continue to work with departments inside and outside UH to provide data science opportunities to the community.

“With the continued support and expertise from Hewlett Packard Enterprise, matched with the curiosity and creative vision of our University of Houston researchers, our place on the leading edge of discovery remains secure,” said HPE DSI Director Claudia Neuhauser in a UH news article.