Md. Sajid Alam Chowdhury

Computer Science PhD Student · Wayne State University

Md. Sajid Alam Chowdhury

Building trustworthy multimodal systems that can reason reliably, preserve privacy, and remain robust.

I am a PhD student in Computer Science at Wayne State University, advised by Prof. Dongxiao Zhu. My research lies at the intersection of multimodal learning and trustworthy AI, with much of my current work centered on Large Vision-Language Models (LVLMs). My current work spans privacy-aware vision-language reasoning, adversarial attacks and defenses, and long-horizon egocentric video understanding and VideoQA, including memory and retrieval mechanisms. Across these directions, I am interested in improving the utility, reliability, and trustworthiness of multimodal systems.

Current focus Privacy-Aware Reasoning
Publications Research Curriculum Vitae

News

Recent Updates

  1. Teaching

    Started as Graduate Teaching Assistant for CSC 3100

    Serving as GTA for Computer Architecture and Organization in the Department of Computer Science at Wayne State University

  2. Publication

    Article published in IET Image Processing.

    Unified adversarial defence framework with diffusion-driven purification, adaptive timestep selection, and reconstruction-consistency detection.

  3. Publication

    Paper published at IEEE QPAIN 2026.

    DDPM-based adversarial purification with noise-guided timestep selection.

  4. Doctoral Study

    Joined Wayne State University and the Trustworthy AI Lab

    Began the Computer Science PhD program and started as Graduate Research Assistant

  5. Industry

    Completed Software Engineer role at Samsung R&D Institute Bangladesh

    Worked on semantic knowledge-graph modeling, reasoning workflows, and applied research before beginning doctoral study.

Show all updates

Research

What I Work On

My research focuses on trustworthy multimodal AI, particularly large vision-language models, with an emphasis on privacy-aware reasoning, adversarial robustness, and long-horizon visual understanding.

Privacy-Aware Vision-Language Reasoning

I study how vision-language models can remain useful while reasoning about privacy-sensitive visual information in a more controlled and reliable way.

  • Privacy boundaries and information disclosure.
  • Privacy leakage in multimodal reasoning.
  • Privacy-utility trade-offs and privacy-aware alignment.

Adversarial Robustness & Security

I study both adversarial attacks and defenses for vision and vision-language models, with a focus on understanding model vulnerabilities and improving robustness.

  • Transferable attacks across models and vision encoders.
  • Adversarial purification and detection.
  • Robustness and generalization under adversarial perturbations.

Long-Horizon Video Understanding

I study how multimodal systems can understand and reason over extended egocentric video histories while maintaining useful and privacy-aware behavior.

  • Long-horizon and egocentric VideoQA.
  • Retrieval and reasoning over extended visual context.
  • Utility and privacy in long-form visual understanding.

Publications

Research Papers

Full Publications

Education

Academic Background

Doctor of Philosophy in Computer Science

Detroit, Michigan, USA

  • Advisor: Prof. Dongxiao Zhu
  • Member of the Trustworthy AI Lab

B.Sc. in Computer Science and Engineering

Chattogram, Bangladesh

  • CGPA: 3.92 / 4.00.
  • Merit Position: 4th out of 127 students, Top 3%.
  • Dean's Award, Faculty of Electrical and Computer Engineering.
  • Advisor: Dr. Kaushik Deb

Higher Secondary School Certificate (HSC)

Dhaka, Bangladesh

  • GPA: 5.00 / 5.00.
  • Served as a Prefect from 2017 to 2019.

Secondary School Certificate (SSC)

Dhaka, Bangladesh

  • GPA: 5.00 / 5.00.
  • Talent Pool Scholarship, 81st position in the SSC examination.

Experience

Work Experience

Wayne State University, Department of Computer Science

Detroit, Michigan, USA

  • Graduate Teaching Assistant for Computer Architecture and Organization (CSC 3100)
  • Supporting instruction, course logistics, student questions, and assessment workflows.

Wayne State University, Trustworthy AI Lab

Detroit, Michigan, USA

  • Conducted research in trustworthy multimodal AI and large vision-language models.
  • Worked under the supervision of Prof. Dongxiao Zhu

Samsung R&D Institute Bangladesh

Dhaka, Bangladesh

  • Contributed to Mantix, a Semantic Knowledge Graph Modeling Tool developed in collaboration with Oxford Semantic Technologies.
  • Built a real-time persistent graph canvas with drag-and-drop support for ontology modeling.
  • Developed a rule-authoring interface for advanced reasoning workflows.
  • Introduced KCP Explorer, a query builder for Samsung's Enterprise Knowledge Cognitive Platform (E-KCP).
  • Contributed to research on real-time adaptive vision correction for VR/AR headsets.

CUET ML Research Group

Chattogram, Bangladesh

  • Conducted research on semantic textual relatedness, multi-dialect Arabic banking intent detection, check-worthiness estimation, multimodal misogynistic meme classification, and offensive language classification.
  • Collaborated with faculty and lab members to design, implement, evaluate, and write machine learning and deep learning research submissions.

Coding

Programming Profile

Solved more than 2000 problems across online judges, with strong competitive programming and software engineering foundations.

Codeforces1607Expert · 1000+ solves · Sajid064 CodeChef20395-Star · 100+ solves · sajid_064 AtCoder9266-Kyu · 300+ solves · Sajid064 LightOJ50+Accepted solves · sajid064

Programming

CC++JavaPythonHTMLCSSSQLJavaScriptSpring BootReactNode.jsNext.jsZustandSPARQLFlutter

ML & Tools

PyTorchTensorFlowKerasscikit-learnOpenCVJupyterGitLaTeXVS CodeRDFox

Awards

Honors & Achievements

International

11th Position, DravidianLangTech@NAACL Task 6: Misogyny Meme Detection.

7th Position, CLEF CheckThat! Lab Task 1: Check-Worthiness Estimation.

7th Position, AraFinNLP Subtask 1: Multi-dialect Intent Detection.

25th Position, SemEval Task 1: Semantic Textual Relatedness.

3rd Position, CodeChef Starters 26 Division 4.

13th Position, CodeChef Starters 33 Division 3.

28th Position, CodeChef Starters 88 Division 2.

Academic & Domestic

Samsung SW Certification Test, Professional Level, first attempt.

Samsung AI/ML Skill Certification, first attempt.

Dean's Award, Faculty of ECE, CUET

Technical Scholarship x8 for academic excellence in CSE, CUET

Honorable Mention x5, ICPC Asia Dhaka Regional Site Preliminary Contest.

99th Position, ICPC Asia Dhaka Regional Onsite Contest.

4th Position, ASRRO ROBOCODER Programming Contest, Round 1.

4th Position, CUET Intra Junior Contest, Round 3.

9th Position, 2nd Runner-up in batch, CUET Intra Junior Contest, Round 2.

5th Position, Champion in batch, CUET Intra Junior Contest, Round 1.

Talent Pool Scholarship, 81st Position, SSC Examination

Extras

Beyond Research

Executive Member, CUET Computer Club

  • Gained competitive programming experience through regular practice and contests.
  • Participated in onsite programming competitions.
  • Helped organize university-level programming contests and club activities.

Prefect, Adamjee Cantonment College

  • Served as a bridge between students and administration.
  • Coordinated assemblies, ceremonies, and institutional events.
  • Supported discipline, communication, and student-facing responsibilities.

Contact

Get in touch

For research collaboration, academic questions, or teaching-related communication, email is the best way to reach me.

sajidalam@wayne.eduPrimary academic email sajid.chowdhury009@gmail.comSecondary email @SajidDijaSTwitter / X GitHub@sajid-alam-chowdhury LinkedInProfessional profile