🖨 Print / Save PDF

Md Anwar Hossen

Ph.D. Student, Computer Science · Iowa State University · Ph.D. Intern, PNNL

Atanasoff Hall, Osborn Drive, Ames, IA 50011 anwarcsejnu@gmail.com +1 (515) 708-7315 linkedin.com/in/anwarcsejnu mahemon.github.io

Career Summary

Graduate student in Computer Science at Iowa State University, specializing in federated continual learning. My research focuses on federated and continual representation learning across heterogeneous data and model architectures by aligning and adapting knowledge representations to improve efficiency and scalability. My work develops representation-level and parameter-efficient methods that fine-tune hidden representations and continually adapt large language models without sharing full models or overwriting prior knowledge. I focus on challenges such as semantic misalignment, domain shift, catastrophic forgetting, and communication constraints to build scalable distributed AI systems.

Education

Ph.D., Computer Science

Iowa State University, USA — CGPA 3.90

M.S., Computer Science

Iowa State University, USA — CGPA 3.89 / 4.00

M.S., Computer Science & Engineering

Jagannath University, Bangladesh — CGPA 3.77

B.Sc., Computer Science & Engineering

Jagannath University, Bangladesh — CGPA 3.71

Research Experience

Ph.D. Intern

Pacific Northwest National Laboratory (PNNL)

  • Research on federated learning for large-scale distributed scientific data, with a focus on heterogeneity-aware representation learning.
  • Designing a Mixture-of-Experts foundation model for domain adaptation and generalization to unseen data domains, without centralizing raw data.
  • Building expert routing and shared representation alignment strategies that transfer knowledge across heterogeneous sources.
  • Preserving privacy, reducing communication overhead, and maintaining robust performance under domain shift.
Ph.D. Intern

Pacific Northwest National Laboratory (PNNL)

  • Developed SCOPE, a communication-efficient framework for federated scientific datasets that leverages intrinsic data structure to handle class imbalance, reduce synchronization and communication overhead, and improve learning efficiency across distributed clients.
M.S. Research

Iowa State University

  • Developed a game-theoretic model for fair bandwidth allocation in hierarchical federated learning systems.
  • Designed distributed heuristics to approximate a Nash equilibrium, improving efficient bandwidth utilization at the edge.
  • Award: Best Poster, Computer Science, Iowa State University (Spring 2023).
Graduate Research Assistant, REACTOR Lab

Iowa State University

  • Built a real-time snowplow navigation system for low-visibility conditions using GPS-RTK lane tracking, radar-based obstacle detection, and a driver interface for operational guidance.

Professional Experience

Assistant Professor (on study leave)

Daffodil International University, Bangladesh

  • Taught undergraduate and graduate courses in AI, Data Structures, and Algorithms; led research and industry collaboration, supervised student research, and organized academic programs.
Development Team Lead

icddr,b, Bangladesh

  • Led development of the Online Doctor's Chamber (ODC) platform, enabling real-time video consultation, appointment scheduling, and secure digital payment for remote healthcare services.
Principal Investigator

National Innovation Fund, ICT Division, Bangladesh

  • Awarded a $3,500 research grant under the Bangladesh ICT Division; designed a system to identify key factors affecting student performance and deployed a framework for real-time assessment.
Software Engineer

MediaSoft Data Systems Ltd, Bangladesh

  • Developed enterprise web applications using ASP.NET MVC and C#, designing modular architectures, implementing routing, and optimizing server-side performance for scalable systems.

Publications

  1. Md Anwar Hossen, Fatema Siddika, Juan Pablo Muñoz, Tanya Roosta, and Ali Jannesari. “CRAFT: Forgetting-Aware Intervention-Based Adaptation for Continual Learning.” Under submission, 2026. (pre-print: arXiv:2605.05732)
  2. Fatema Siddika, Md Anwar Hossen, Tanwi Mallick, and Ali Jannesari. “Sparse Subspace-to-Expert Sharing for Task-Agnostic Continual Learning.” Under submission, 2026. (pre-print: arXiv:2606.07500)
  3. Md Anwar Hossen, Nathan Tallent, Luanzheng Guo, and Ali Jannesari. “SCOPE: Semantic Coresets using Orthogonal Projection Embeddings for Federated Learning.” Under submission, 2026. (pre-print: arXiv:2603.12976)
  4. Md Anwar Hossen*, Fatema Siddika*, Juan Pablo Muñoz, Tanya Roosta, Anuj Sharma, and Ali Jannesari. “FedReFT: Federated Representation Fine-Tuning with All-But-Me Aggregation.” 19th Conf. of the European Chapter of the ACL (EACL), 2026. (*Equal contribution)
  5. Fatema Siddika, Md Anwar Hossen, Wensheng Zhang, Anuj Sharma, Juan Pablo Muñoz, and Ali Jannesari. “Dual-Distilled Heterogeneous Federated Learning with Adaptive Margins for Trainable Global Prototypes.” 26th IEEE Int. Symp. on Cluster, Cloud, and Internet Computing (CCGrid), 2026. (pre-print: arXiv:2508.19009)
  6. Md Anwar Hossen, Fatema Siddika, and Wensheng Zhang. “Fair Allocation of Bandwidth at Edge Servers for Concurrent Federated Learning Processes.” 10th Int. Conf. on Fog and Mobile Edge Computing (FMEC), 2025.
  7. Md Anwar Hossen, Rakib Bin Alamgir, Arman Ul Alam, Fatema Siddika, Shah Fahad Hossain, and Md. Shohel Arman. “A Web-Based Four-Tier Architecture using Reduced Feature-Based Neural Network Approach for Prediction of Student Performance.” Int. Conf. on Robotics, Electrical and Signal Processing Techniques (ICREST), 2021.
  8. Md Anwar Hossen, Emran Hossain, Abdul Khalib Zahereel Ishwar, and Fatema Siddika. “Ensemble Method-Based Architecture using Random Forest Importance to Predict Employee Turnover.” Journal of Physics: Conference Series, 2021.
  9. Md Anwar Hossen, Md. Shariful Islam, Nurhafizah Abu Talip Yusof, Md. Sakib Rahman, Fatema Siddika, Mostafijur Rahman, Sabira Khatun, Mohamad Shaiful Abdul Karim, and S. M. Hasan Mahmud. “Hybrid Sampling and Random Forest-Based Machine Learning Approach for Software Defect Prediction.” 5th Int. Conf. on Electrical, Control & Computer Engineering, 2019.
  10. Md Anwar Hossen, Fatema Siddika, Tonmoy Kumar Chanda, and Touhid Bhuiyan. “A Comparison of Soft Computing Methods on Imbalanced Data.” Int. Conf. on Cyber Security and Computer Science (ICONCS), 2018.
  11. Fatema Siddika, Md Anwar Hossen, and Sajeeb Saha. “Transition from IPv4 to IPv6 in Bangladesh: A Competent and Enhanced Approach.” 2nd Int. Conf. on Networking Systems and Security (NSysS), 2017.

Honors & Awards

  • Best Research Poster, Computer Science, Iowa State University — 2023
  • National Innovation Fund Grant (worth $3,500), ICT Division, Bangladesh — 2019–2020
  • National Science and Technology Research Fellowship for MS, Bangladesh — 2016–2017
  • Research Fellow, BdREN Summit, University Grants Commission, Bangladesh — 2016
  • Honorable Mention, ACM ICPC Asia Regional Dhaka Site — 2013
  • Runner-up, Code Warrior Challenge (National level), BASIS & ICT Division — 2014
  • Reached Top 15, Therap Java Fest, Bangladesh — 2014

Professional Service

  • Lead, Technical Session Organizing Committee, 2nd Int. Conf. on Cyber Security and Computer Science (DIU & Karabük University, Turkey) — 2020
  • Administrative Coordinator, World Friends IT Volunteer Program, South Korea — 2019
  • Mentor, IEEE Daffodil International University Student Branch — 2017–2019
  • Chair, IEEE Jagannath University Student Branch — 2017
  • Mentor and Judge, National Hackathon, ICT Division, Bangladesh — 2016

Teaching

Instructor · Assistant Professor

Daffodil International University, Bangladesh

  • Data Structures, Computer Algorithms, Discrete Mathematics, Artificial Intelligence.
Teaching Assistant

ComS 228: Data Structures

Certifications

  • Federated Fine-Tuning of LLMs with Private Data — DeepLearning.AI
  • Probability and Statistics for Machine Learning and Data Science — Coursera
  • Neural Networks and Deep Learning — Coursera
  • Improving Deep Neural Networks: Tuning, Regularization, and Optimization — Coursera
  • Build Basic Generative Adversarial Networks — Coursera