Skip to content
View razon1494's full-sized avatar
2 followers · 0 following

Achievements

Achievement: YOLO

Achievements

Achievement: YOLO

Block or report razon1494

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
razon1494/README.md

Hi 👋, I'm Mohammad Arifur Rahman

M.S. Computer Science | Trustworthy NLP & LLM Evaluation

Portfolio · rahman.arif.cse@gmail.com

About

  • M.S. Computer Science student at Anderson University, based in Huntsville, Alabama.
  • I study the reliability of AI-generated-text detectors under paraphrasing and distribution shift.
  • I am preparing this research for academic submission and pursuing doctoral research in trustworthy NLP.

Research focus

Trustworthy NLP LLM Evaluation AI-Text Detection Robustness and Distribution Shift

Featured work

LLM-text detection under iterative paraphrasing

A reproducible evaluation of character-TF-IDF, word-TF-IDF, and RoBERTa detectors under iterative standard and simplified paraphrasing.

Key result: On a readability-defined hard subset, character-TF-IDF F1 decreases from 0.923 to 0.336 after two simplified paraphrasing rounds.

Toolkit

Python · PyTorch · Hugging Face Transformers · scikit-learn · pandas · NumPy · Jupyter · LaTeX · Git

Portfolio: ari-fur.com · Email: rahman.arif.cse@gmail.com

Pinned Loading

  1. Analysis-of-Human-vs.-LLM-Generated-Text Analysis-of-Human-vs.-LLM-Generated-Text Public

    Reproducible study of hardness-stratified LLM-text detector robustness under iterative paraphrasing across TF-IDF, RoBERTa, and three paraphrasers.

    Python

  2. membership-inference-probing-nlp membership-inference-probing-nlp Public

    Investigating whether iterative paraphrase perturbation erases training data membership signals in language models — behavioral probe analysis across perplexity, loss curvature, zlib ratio, and low…

    Jupyter Notebook

  3. stress-test stress-test Public

    eliability evaluation framework for AI-text detectors — hardness-aware, semantics-controlled, symmetric false-accusation analysis under real-world transformations.

    Python

  4. clip-prompt-domain-shift clip-prompt-domain-shift Public

    Evaluating prompt learning (CoOp) vs. zero-shot and linear-probe CLIP under domain shift on PACS and OfficeHome

    Python

  5. sql-query-feedback-system sql-query-feedback-system Public

    Automated SQL query evaluation and feedback system with AST comparison, provenance tracing, and misconception detection. PhD research prototype — Mohammad Arifur Rahman | Dr. Hasan Jamil, Universit…

    Python

  6. ECG-Arrhythmia-Temporal-Attention ECG-Arrhythmia-Temporal-Attention Public

    ECG arrhythmia detection using Multi-Scale Temporal CNN + Pre-LayerNorm Transformer. Handles severe class imbalance (MIT-BIH, 5 AAMI classes) with focal loss, SE attention, mixed-precision training…

    Python