20th August 2022

ISSAI Summer Research Program 2022 interns share their experiences

This year, the ISSAI team has again organized a summer research program for students from various Kazakhstani universities. The program offers students the opportunity to gain knowledge in artificial intelligence, machine learning, programming and related fields by working closely with ISSAI data scientists. During the program, students are assigned a mentor from the ISSAI data science team with whom they work on real-life AI projects. 

The Summer Research Program is an initiative of the ISSAI team that first took place in the summer of 2021 and has been held every summer since. The main goal of the program is to give students from all over Kazakhstan the opportunity to gain hands-on experience working on AI projects, acquire knowledge and skills in the field of AI, and gain real-world experience working with data scientists. The program lasts approximately two months during the summer, when students typically have their summer vacation. 

To be admitted, applicants must go through a selection process. The students with the best results and academic performance are then selected. This year, we received applications from universities, such as Eurasian National University, Kazakh British Technical University, Suleyman Demirel University, Yale University, Nazarbayev University, Skolkovo Institute of Technology, Payame Noor University, Al-Farabi Kazakh National University and others.

We conducted interviews with three Summer Research Program’22 (SRP) graduates who volunteered to provide their feedback: Moldir Berkaliyeva, Zhandos Ayupov, Arman Bolatov, Aidar Amangeldi, and Kaisar Dauletbek.

1.    Hello! Thank you for applying for the SRP! Please tell us why you originally decided to apply. What was the main reason?

Moldir: The main reason was to put my knowledge into practice, gain experience, and work on real projects.

Zhandos: I wanted to gain experience with machine learning, especially deep learning. So, I thought the SRP was a good option.

Arman: I wanted to help ISSAI with some of their ongoing projects and improve my knowledge in computer vision and natural language processing.

Aidar: I found out about the SRP through ISSAI’s Instagram page. The reason why I applied for this program is because I wanted to gain research and programming experience in Data Science.

Kaisar:  I wanted to somehow contribute to the development of Kazakh speech technologies, and I knew that people at ISSAI were doing that. I already had some background in natural language processing, so I decided to try my hand at speech recognition. I also knew that the computational resources at ISSAI were immense, and I was eager to play with the best tech available in the world. 

2.    What is your background in terms of university major/specialization?

Moldir: Mathematics.

Zhandos: Information technologies at the Kazakh-British Technical University.

Arman: Computer Science.

Aidar: Computer Science.

Kaisar: Mathematics.

3.    Now that the Program is coming to an end, are you happy with the results, with the things you’ve learned?

Moldir: I’m more than happy and grateful to have attended the SRP. Seeing the results of your work in the form of correctly working codes and the knowledge I have gained is the best reward.

Zhandos: Yes, I am very happy with the results.

Arman: Yes, I’m happy!

AidarYes, in the two months of the program I’ve gained a lot of valuable hard and soft skills. I have learned how to scrape data from the web and label images, use different software tools, and read research papers. I’ve also tried different state-of-the-art machine learning architectures. Speaking of soft skills, I strengthened my communication and collaboration skills by working in a team. I believe it was fruitful work for our team.

KaisarYes, I am satisfied with the results. I would be even happier if we could publish the paper, but we will probably publish it later, so everything is good.

4.    What was the project you worked on? And who was your supervisor?

Moldir: Kazakh Virtual Assistant for People with Visual Impairment. Supervisor – Askat Kuzdeuov.

ZhandosMy supervisor was Aknur Karabay. I worked on two projects: The Use of Deep Learning Methods to Predict the Optimal Structure of Reinforced Columns and Food Recognition for the Central Asian Food Dataset, which also required data collection and labeling.

Arman I worked on the creation of the Central Asian food dataset and its classification. My supervisor was Aknur Karabay.

Aidar: I worked on the food recognition team led by Aknur Karabay. 

Kaisar: I worked on Multilingual Speech Recognition For Turkic Languages. My supervisor was Saida Mussakhojayeva.

5.    What were some of the tasks you had to do while working on this project?

Moldir: Every week, my teammates and I were given several tasks by our supervisor. First, I watched tutorials on optical character recognition (OCR) and familiarized myself with this term by implementing existing codes. Then, I was given tasks to apply similar algorithms to our project, involving the Kazakh language, mainly, reading book pages from different perspectives and decoding bar/QR codes.

Zhandos:  I learned to collect, search, and label data, develop models, and fine-tune hyperparameters. I also read research papers related to my project area.

Arman:  I scraped Google and Yandex for food images. I created a project in Roboflow and labeled the data for object localization. Then, I built a cropping and resizing script to create a classification dataset.

Aidar: I had to retrieve images of popular Central Asian dishes (e.g. besbarmaq, qazy, manty, etc.) from social media platforms, such as Facebook, Instagram, VK. I also had to annotate images and remove irrelevant ones.

Kaisar: I did a lot of data preparation and model training. Tasks usually included preparing dataset statistics, conducting specific experiments, and literature review. 

6.    Did your supervisor help you along the way? If yes, how?

Moldir: My supervisor supported and guided me all the way through. We had weekly meetings to discuss and show the results of the assigned tasks. In general, whenever I had any questions regarding the tasks, I could send him a message or ask him in person. I am very grateful to my supervisor for that.

Zhandos: Yes, we had meetings where we shared progress and solved problems. Also, we were in touch via messengers.

Arman: My supervisor guided me throughout the SRP and helped me a lot. We had 2-3 meetings every week and communicated daily in a Telegram chat.

AidarYes, we had weekly meetings and communicated daily through a Telegram chat, where the supervisor assigned tasks, tracked our progress, gave tips, and answered questions. 

KaisarYes, Saida was extremely helpful, she guided me through pretty much every question I had. We had three meetings a week during the SRP, and we usually texted each other any time something came up.

7.     Did you gain any new knowledge, or expand your knowledge, in a particular area while working on the project ?

Moldir: Yes, previously I was familiar with the main machine Learning algorithms in general. During the program, I had a chance to study and develop a Kazakh reading assistant using OCR algorithms.

Zhandos: Yes. For example, I gained new knowledge about computer vision and learned about tools like Docker.

Arman: Yes! I explored current SOTA methods for object detection and image classification, such as YOLO, CoCo, ViT, EfficientNet. I have improved my knowledge in computer vision.

AidarI enhanced my knowledge in Data Science, especially in object recognition/localization.

Kaisar: Yes. I learned a great deal about speech recognition (latest methods, literature, frameworks), Linux, and everything that fell in between the tasks I received.

8.    Do you think the SRP was beneficial for you? What did you like about this program? 

Moldir: For sure, yes. Besides the content of my work, I liked it that the program was well-organized right from the recruitment process. The requirements were clearly described on the recruitment poster. During the interview, the ISSAI members answered all of my questions regarding the workflow. My supervisor gave me tasks at the beginning of each week and tracked my progress.

Zhandos: Sure, I gained a lot of experience in the research area. And I think it will help me to build my career.  It was a very good experience, and I am going to write an article about my experience in my personal blog.

Arman: Yes, it was beneficial. The program showed me what research assistants basically do and how to work in academia.

AidarIt was very useful for me. I liked the topic of the team I worked in.  I am happy to have had such an experience, because it was my first internship experience where I gained real industry and research skills.

Kaisar: I do think the SRP was beneficial for me. I liked how well-organized everything was – regular meetings, clear tasks and deadlines, and the uniqueness of the project we worked on.

9. Would you recommend the SRP to your fellow students and friends?

Moldir: Yes, I would definitely recommend the ISSAI SRP as a way to have an interesting and productive summer.

Zhandos: Yes, sure. I already recommend the SRP to potential candidates for next year.

Arman: Yes.

AidarDefinitely, yes.

Kaisar: Yes, absolutely, but only for those who I know can work hard and contribute properly.