Go to Diversity in HPC home page

Faces of HPC: Jesmin Jahan Tithi

Jesmin Jahan Tithi is an HPC Software Architect at the Platform Architecture and Performance Team at Intel Corporation. Jesmin previously completed a Ph.D. on “Engineering and high performance parallel algorithms with applications to bioinformatics” at Stony Brook University, New York, USA.


Jesmin Jahan Tithi went against the wishes of her family to study Engineering in Bangladesh. Jesmin’s father wanted her to accept a medical school admission and she also faced discrimination as one of only a few women in her class. However, Jesmin has now completed a Ph.D in HPC and its applications and works as an HPC Software Architect at Intel.


Tell a bit about yourself

I am from Bangladesh, the 8th most populated country in the world but rank 97th in terms of land area. I received my B.Sc. in Computer Science and Engineering from the Bangladesh University of Engineering and Technology (BUET), which is the top-ranked engineering school in Bangladesh. I received my Ph.D. from the Stony Brook University on “Engineering high-performance parallel algorithms with applications to bioinformatics.” I like to post my songs on youtube, watching animation movies and traveling.

What is your current job?

I am working as an HPC Software Architect at the Platform Architecture and Performance Team at Intel Corporation, Santa Clara, CA, USA. I work on developing, implementing, optimizing new parallel algorithms for solving computational problems on a variety of parallel architectures. During my Ph.D., I worked on engineering high-performance algorithms with applications to bioinformatics. I worked on providing algorithmic improvements to make traditional algorithms more parallel, scalable, cache-, space-, bandwidth- and energy-efficient. At Intel Corporation, I am mainly working on machine learning algorithms, next-gen programming platforms and also in bioinformatics.  

While I was choosing my Ph.D. topic, my first choice was networking and sensors. My second preference was bioinformatics. However, after coming to Stony Brook, I was not able to find a faculty member who works in sensor networks with funding for a new student. I took Natural Language processing and machine learning courses which I found very interesting and intuitive to me. But again, the faculty who was interested in taking me could not guarantee any funding. At the same time, another professor was looking for new students to work in parallel algorithms and bioinformatics. So, I jumped into to grab that opportunity. I had no prior experience in parallel programming or high-performance computing. I learned it from scratch.

How did you become interested in HPC?

It was through my Ph.D. research I became interested in HPC. It was really a fun to see how the runtime of a program can be decreased gradually by increasing number of cores, how tough debugging a parallel program is, why something works and some doesn’t, how one can reason about different performance trends on parallel machine, how to run a program on thousands of distributed machines, and the feeling that “HPC is everywhere” and “HPC transforms”.

I started my HPC education in a grad course as part of my Ph.D. at 2012. I was able to publish a poster and one lighting talk just after 6 months of work at SC 12 and that was an eye-opening conference for me, exposing the world of HPC and opportunities. From the same course work, I published two more papers at IPDPSW (International Parallel & Distributed Processing Symposium and workshops). In 2013, I worked as a hardware engineering Intern at Intel Corporation, at Hudson with some amazingly talented people, who inspired me a lot. I kept attending and presenting my work at many HPC and bioinformatics conferences, figuring out what are all the options I can contribute to. I worked in Google Inc as a software engineering intern in the Flume and Data Flow Engine and I worked on the backend of this to provide load-balancing and quality of service on a distributed settings. Furthermore, I interned in the Pacific Northwest National Lab HPC team to work in devising a method of predicting performance of a parallel application. Finally, after my graduation, now I am working as an HPC software architect at Intel Corporation. And my journey continues …

Is there something about you that’s given you a unique or creative approach to what you do?

I believe in deep of my heart that doing my work with full responsibility is also a form of worshiping Almighty. That helps me to be productive.

Were there any challenges when you first entered the field?

The challenge started from my undergrad. My father wanted me to accept a medical school admission offer instead of the engineering school offer since he used to think that it is inconvenient for women to work as an engineer and there are very few opportunities for women as engineers. I had to go against his wishes to study in the engineering school.

I used to face difficulties due to the dominance of male students in the class, lab courses or even in getting assistance of teachers due to being female. Even during my Ph.D. studies, I had a male professor or a colleague at the Corporation commenting that I am getting special benefits since I am female. A male professor said he envies me for the privileges/benefits I am receiving which I felt very disturbing. In most of the HPC conferences fewer attendees are female. Since, often I am the only girl in a room, it is also intimidating.

What’s the best thing about working in HPC?

Working in HPC is fun and challenging, with lot of opportunities to contribute in applications that matter!

If there’s one thing about HPC you could change, what would it be?

There is a lot of bloat in performance numbers shown. People should understand the theoretical implications of performance scalability instead of blindly going for performance that stops scaling after some point and waste resources.

What’s next for you in HPC ?

I see myself influencing next generation computational hardware, programming platform, software design, and algorithmic choices in Machine Learning and Bioinformatics.

Last updated: 11 Jul 2016 at 12:07