1 of 44

UGRIP: Undergraduate Research Internship at MBZUAI

​

NAZARBAYEV UNIVERSITY, OCTOBER 28

2 of 44

UGRIP 2025

Damir Kim

4th-year Mathematics student

Moldir Azhimukhanbet

4th-year CS student

Artur Muratov

4th-year CS student

3 of 44

Today’s Talk

Overview of the UGRIP Program

Research in Natural Language Processing — Artur

1.

2.

Research in Computer Architecture — Moldir

3.

Research in Computer Vision — Damir

4.

4 of 44

UGRIP Program

1.

5 of 44

UGRIP Program

  1. Fully funded, four-week Research Internship at MBZUAI (Abu Dhabi).
  2. In total 15 projects: CS, CV, ML, NLP, and Robotics etc.

6 of 44

Projects 2025

1. Brain Tumor Segmentation

2. FRaN-X: FRaming & Narratives-eXplorer

3. Automated End-To-End Media Bias Rating System

4. Certified Learning for Safe Obstacle Avoidance in Robotics

5. Enhancing Uncertainty Quantification for Large Language Models through Chain-of-Thought Reasoning on Question Answering Tasks

6. ML System Optimization for Multi-Modal and Interactive Music Generation

7. Toward Role-Aware Large Language Models for Safe Organizational Deployment

8. Early Explorations on the Impact of Pruning and Low-Bit Quantization for LLM Reasoning Ability

9. Exploring Jailbreak Attack Transferability Between Speech Language Models

10. Rethinking Tokenization and Embedding in Language Models

11. Tokenization-free Machine Translation for Dialectal Arabic

12. Improving LLM performance for code-switching data: a new multitask benchmark and an investigation into cross-lingual alignment

13. Curriculum-learning for cross-lingual knowledge distillation

14. Asm-to-Asm Transpilation

15. Project Telepathy: Building an AI Professor X

7 of 44

Why UGRIP?

  • To experience MBZUAI and Abu Dhabi.
  • To have a publication.
  • To boost your chances for a Master’s offer.
  • To understand if research is for you.

+:

Many GPUs

Free coffee/tea

Single dorm rooms

​

-:

Small medical center

Very hot in summer

30 min. from center

8 of 44

MBZUAI and Abu Dhabi

9 of 44

Application

  • UGRIP 2025: 5 out 57 (9%) from NU.
  • Will open on January 1, 2026 (Rolling admissions —> apply early).
  • Minimum CGPA of 3.5 (on a 4.0 scale) —> high GPA is very important.
  • Required documents (request before the end of Fall semester):
    1. Publications/ Research Experience —> big advantage. Where? ISSAI
    2. Official transcript (in application math & cs grades listed separately).
    3. Letter from University registrar confirming English (instead of IELTS).
    4. Statement of purpose essay (MBZUAI Tech Talks).

10 of 44

Once you are accepted:

  • If you want to buy AED, do it in Astana.

​

  • Project Allocation:�
    • You will be asked to choose Top 3 projects (at MBZUAI).
    • Deallocation system tries to place everyone into one of their choices =>

Important tip: If there’s a highly selective project you really want, don’t put “unpopular” projects as #2 and #3, this can misplace you

(I made this mistake).

​

  • You will be evaluated.

11 of 44

CS/CA Research

2.

12 of 44

Background

Moldir Azhimukhanbet�Senior Year Computer Science Student

​

  • 3 pet projects (one with FPGA)�
  • Competitive GPA
  • Texer.AI - Hardware Design and Verification Engineer�

13 of 44

Asm-to-Asm Transpiler

14 of 44

Asm-to-Asm Transpiler

15 of 44

Asm-to-Asm Transpiler for o2 optimization level

16 of 44

Asm-to-Asm Transpiler for o2 optimization level

17 of 44

Asm-to-Asm Transpiler

18 of 44

Asm-to-Asm Transpiler

19 of 44

Asm-to-Asm Transpiler

20 of 44

Asm-to-Asm Transpiler

21 of 44

My Professor

22 of 44

CV Research

3.

23 of 44

Background

  • 1 conference experience in India�
  • RA at NUSOM, CAFI Lab�
  • Internship @ISSAI, SRP 2024

​

  • 2-3 pet projects�
  • 1 Coursera certificate�
  • Competitive GPA (>3.5)�

24 of 44

Brain Tumor Segmentation, Generation, Classification

Three tasks from BraTS 2025 Lighthouse Challenge

� Task №11: Predicting the Tumor Response During Therapy� Task №8: MR Image Synthesis�

Task №5: Segmentation

…state-of-the art methods can be widely implemented in Sub-Saharan Africa given the extensive use of lower quality MRI technology, which produces poorer image contrast and resolution…

25 of 44

Segmentation, Generation, Classification Teams

26 of 44

Project Supervisor

Dr. Mohammad Yaqub�Associate Professor of Computer Vision

​

  • Сompleted his PhD from the University of Oxford in 2011

​

  • University lecturer at IT services, University of Oxford and a research fellow at the Nuffield department of clinical neurosciences (NDCN), University of Oxford

​

  • Founder of Labib AI, MBZUAI first startup

​

  • Founder of BioMedia Lab

27 of 44

Brain Tumor Segmentation Task

28 of 44

Brain Tumor Segmentation Task

Need to classify these:

​

​

​

​

​

​

​

​

All we have: 60 annotated cases MRI scans from the dataset from Sub-Saharan Africa, gathered from Nigeria, possibly from the single hospital.

29 of 44

What did we use?

30 of 44

What did we use?

31 of 44

A lot of troubleshooting & training…

32 of 44

My first predictions:

33 of 44

Did we do well?

34 of 44

Results

35 of 44

Awards

Best Team Award

3rd Place in the Challenge

36 of 44

NLP Research

4.

37 of 44

Background

Artur Muratov�

​

  • 2 month RA at @ISSAI, (1 project)

​

  • Gen-AI project (3rd place at startup competition in Doha, Qatar; 3rd Place nFactorial Incubator Demo Day)

​

  • CGPA 3.92/4.00, ranking #1

​

  • 2nd place at Informatics&Physics at Phystech.International Olympiad (high school)

38 of 44

FRaN-X: Framing and Narratives-Explorer

  • Goal: Detect how people, countries, or organizations are portrayed in news articles.
  • Part of SemEval-2025 Task 10.

39 of 44

SemEval-2025 Task 10

40 of 44

SemEval-2025 Task 10

41 of 44

SemEval – What Is It?

The top global benchmark competition in NLP

​

​

​

​

​

​

​

​

42 of 44

My Professor

Preslav Nakov

Department Chair and Professor of Natural Language Processing

43 of 44

Q&A

5.

44 of 44

mbzuai.ac.ae

Mohamed bin Zayed University of Artificial Intelligence

Masdar City, Abu Dhabi, United Arab Emirates

​

Thank you