“An investment in knowledge pays the best interest.”

The learning curve has a beginning but not an end. Investing in young talent has a great impact on a nation because the future depends on the youth, especially in a country like Lebanon. To help Lebanese students gain global knowledge and prepare them for the labor market, Maroun Semaan’s Faculty of Engineering and Architecture (MSFEA) at The American University of Beirut (AUB) partnered with LebNet to give engineering graduates the rare opportunity to work with US-based companies for their final year projects (FYPs) and course projects. 

Four teams from AUB worked with two companies in the US: Asurion (a device insurance, warranty, and support services provider for cell phones, consumer electronics, and home appliances) and FADEL (the creator of rights and royalty management software). 

In 2019 and 2020, each of the four teams either worked on an FYP or a course project, closely collaborating with mentors from FADEL and Asurion.

(In collaboration with LebNet, four teams from Maroun Semaan Faculty of Engineering and Architecture at AUB worked on course and final year projects with US-based companies Asurion and Fadel.)

Projects at a Glance 

Expert Helper (FYP)

Students: Sara Hammoud, Aya Eido, and Dana Daoud
Company: Asurion
Mentors: Peng Xie and Sundar Kuppuswamy

The Work:

The AUB team prepared and curated a data set of tech support sessions from Asurion into a category of replies. They then selected categories of these sets and automated initial replies to them using natural language processing techniques mixed with expert knowledge. 

(Aya Eido, Sara Hammoud, and Dana Daoud)

The Experience:

“The collaboration went very well. Our goal was to make sure that students worked on a problem that interests them and has a potential value for Asurion and to ensure the team learned from a tech standpoint how to implement new algorithms and solve problems at hand while getting a sense of data in the real world,” said Sundar Kuppuswamy. “My experience with the students was good. The students were very curious and motivated, did a great job exploring the original data set, and came up with multiple ideas. I would be happy to repeat the exercise next year between AUB and the team at Asurion,” added Peng Xie. 

“The last academic year was definitely not easy. Our team had to adapt to many challenges and work hard to be able to deliver what we promised, all while taking care of our well-being and mental health. If it weren’t for my teammates and the culture we established that is based on openness, optimism, and trust, we would not have been able to submit the requirements, let alone be nominated for the Murex Best Innovative Software Development Award and present our work to Asurion’s Chief Analytics Officer, Faker Zouaoui,” revealed Sara Hammoud. 

Emerging Problems (Natural Language Processing NLP Course Project)

 Students: Julia Zini and Issa Issa
Company: Asurion
Mentor: Peng Xie and Sundar Kuppuswamy

The Work:

The AUB team helped Asurion’s tech support team figure out whether a novel tech problem is emerging on social media (Twitter). And for novel emerging problems, it also helps determine if the problem is related to tech support or generic news events.

(Julia Zini and Issa Issa)

The Experience:

“Working with industries on a course project gives you a different perspective, because usually most of the university projects are research-oriented and not backed by delivery. It was especially interesting for me and Julia because we had to deliver a well-packaged product and the insights from Asurion and feedback were rewarding,” commented Issa Issa. 

Extracting Insights (NLP course project)

Students: Mohamad Mansour, Fouad Khnaiser and Bassel Musharrafieh
Company: Asurion
Mentor: Harsh Tomar and Sundar Kuppuswamy

The Work:

The AUB team focused on extracting phrases of trends from collection of text data (emails) allowing the Customer Experience team to quickly identify and mitigate issues. The project was hosted by a startup incubated by Asurion, which had different IP regulations. This prohibited Asurion from sharing the data as they discovered they required different NDAs. Despite this, the team worked on a methodology to extract information from public data similar to what Asurion might have. The results were impressive enough to be accepted as a possible solution. It was a learning curve for the team because they had to apply NLP techniques to an industry-level problem and deal with real-data. 

The Experience: 

“The industry project provided a great opportunity for the students to experiment with real business problems. Where in an academic problem, students try to solve problems to get to the right answer via the right methods, in business problems, there sometimes isn’t a right answer, and oftentimes no “right method”.
The students broke the problem down into smaller pieces and attacked each piece sequentially with the easiest methods to get the outcomes. At each step, new problems emerged and so did several different ideas to solve them. Key steps from the students’ implementation of phrase extraction ended up being utilized in the working of the ‘Extract Insights’ project.” – said Harsh Tomar.

Image Match (FYP)

Students: Hadi Ahmad, Hafez Jawhary, and Samir Saidi
Company: FADEL
Mentors: Rony Eid and Ziad Bassil

The Work:

Specialized in copyright and digital rights management, FADEL’s goal is to ensure that the digital content its clients use does not violate any copyright laws. Hence, the AUB team was tasked with improving matching performance. Walid Daccache, FADEL’s CTO, explained that with the help of FADEL mentors, who met with students on a weekly basis, the team implemented a different algorithm that outperforms FADEL’s algorithm while being compatible with the rest of their system. 

(Hadi Ahmad, Hafez Jawhary, and Samir Saidi)

The Experience: 

“My colleagues and I agreed with FADEL to extend this project beyond the course’s frame. The complexity and time requirements of our assignment ensued this mutual understanding over the project’s time management. The new image detection model is substantially accurate for large datasets, while still maintaining adequate performance,” shared Samir Saidi. 

“We were glad to work with the AUB faculty members and their bright students on finding solutions to challenging problems in image processing. We feel that the collaboration and knowledge exchange between FADEL engineers and the AUB FYP team added value to all parties who participated in the project,” said Daccache. 

“Though the final solution required some refinement in terms of accuracy and performance but still Samir Saidi, one member of the team, continued to work on it within his internship with the company and that added additional plus points to the solution towards its feasibility to be integrated within our product. Eventually the collaboration yielded good results on which we can build further to reach more successes,” said Rony Eid. 

For the Future

Lebanon is suffering from many crises and significant challenges are facing the education sector and students. But such collaborations bring hope for a better generation and future.