Anwar Zahid

Researcher & engineer Iowa State University

Finding the fault lines in intelligent systems.

I’m Anwar Zahid. I study how AI and numerical software fail, then build tools to make those failures easier to detect, reproduce, and fix.

AI reliability Numerical debugging Software engineering

Anwar Hossain Zahid Portrait of Anwar Hossain Zahid
01 / PROFILE Anwar Hossain Zahid
Exploring reliability beneath the surface Scroll to discover

## News · recent log

News

## About

From shipping software to studying why it fails.

I am a Ph.D. student in Computer Science at Iowa State University, advised by Prof. Wei Le in the Program Analysis and AI Lab. My research focuses on reliable AI systems, numerical debugging, and software engineering techniques for machine learning systems.

Before starting my Ph.D., I spent four years as a software engineer on banking, e-governance, AI, and mobile platforms. That industry background shapes how I approach research: I care about methods that can become practical tools for developers and researchers. A fuller timeline is on the extended about page.

## Research interests

What happens beneath the surface?

Can we trust the answer?

Studying LLM hallucination and evaluating where intelligent systems lose their reliability.

Where does a number go wrong?

Tracing numerical instability and silent regressions through the software that powers machine learning.

How do we make it useful?

Building interpretable testing and debugging tools that help people understand and fix failures.

## Publications

Peer-reviewed & preprint work

On ML reliability, LLM evaluation, and numerical correctness.

ML reliability Soft Assertions Find the bugs hidden in the numbers.
FSE 2025 / PACMSE

Automatically Detecting Numerical Instability in Machine Learning Applications via Soft Assertions

Shaila Sharmin, Anwar Hossain Zahid, Subhankar Bhattacharjee, Chiamaka Igwilo, Miryung Kim, and Wei Le

Introduces Soft Assertions, a method for detecting and triggering hidden numerical instability bugs in machine learning applications.

Paper Code
BibTeX 
@article{sharmin2025softassertions,
  title   = {Automatically Detecting Numerical Instability in Machine Learning Applications via Soft Assertions},
  author  = {Sharmin, Shaila and Zahid, Anwar Hossain and Bhattacharjee, Subhankar and Igwilo, Chiamaka and Kim, Miryung and Le, Wei},
  journal = {Proceedings of the ACM on Software Engineering (PACMSE), issue FSE 2025},
  year    = {2025},
  doi     = {10.1145/3729394},
  url     = {https://arxiv.org/pdf/2504.15507}
}
LLM evaluation Context matters. Hate speech detection across geographies.
arXiv:2502.19612

Evaluation of Hate Speech Detection Using Large Language Models and Geographical Contextualization

Anwar Hossain Zahid, Monoshi Kumar Roy, and Swarna Das

Evaluates how large language models perform on hate speech detection when geographic and social context are included.

Paper
BibTeX 
@misc{zahid2025hatespeech,
  title         = {Evaluation of Hate Speech Detection Using Large Language Models and Geographical Contextualization},
  author        = {Zahid, Anwar Hossain and Roy, Monoshi Kumar and Das, Swarna},
  year          = {2025},
  eprint        = {2502.19612},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2502.19612}
}
GPU numerics NVIDIA / AMD Same computation. Different results.
arXiv:2410.09172

Testing GPU Numerics: Finding Numerical Differences Between NVIDIA and AMD GPUs

Anwar Hossain Zahid, Ignacio Laguna, Wei Le

Studies numerical differences between NVIDIA and AMD GPU executions and their implications for reproducibility and portability.

Paper
BibTeX 
@misc{zahid2024gpunumerics,
  title         = {Testing GPU Numerics: Finding Numerical Differences Between NVIDIA and AMD GPUs},
  author        = {Zahid, Anwar Hossain and Laguna, Ignacio and Le, Wei},
  year          = {2024},
  eprint        = {2410.09172},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2410.09172}
}
Software engineering education Learn by building. A conceptual platform for virtual internships.
ICCIT 2020

A Conceptual Design of Virtual Internship System to Benchmark Software Development Skills in a Blended Learning Environment

Raihan Mia, Anwar Hossain Zahid, Bipul Chandra Dev Nath, and Abu Sayed Md Latiful Hoque

Presents a conceptual software platform for virtual internship delivery and software development skill benchmarking.

Paper
BibTeX 
@inproceedings{mia2020virtualinternship,
  title     = {A Conceptual Design of Virtual Internship System to Benchmark Software Development Skills in a Blended Learning Environment},
  author    = {Mia, Raihan and Zahid, Anwar Hossain and Nath, Bipul Chandra Dev and Hoque, Abu Sayed Md Latiful},
  booktitle = {2020 23rd International Conference on Computer and Information Technology (ICCIT)},
  year      = {2020},
  publisher = {IEEE},
  doi       = {10.1109/ICCIT51783.2020.9392670},
  url       = {https://ieeexplore.ieee.org/document/9392670}
}

## Experience

Research grounded in real-world software.

Experience across research, banking, e-governance, and mobile software.

## Projects

Selected systems & applied work

Coursework and smaller repositories are collapsed below.

AI literacy Today’s AI Discover tools. Learn how to use them.

Today's AI — Find AI tools and learn how to use them

A web platform that helps K-12, undergraduate, and graduate learners — and anyone curious about AI — discover the right AI tools and learn to use them, built as Creator & Lead Developer under the direction of Prof. Wei Le.

  • Web platform
  • LLM assistant
  • Search
  • AI literacy
Demo
ML testing Soft Assertions Detect numerical instability in ML.

Automatically Detecting Numerical Instability in Machine Learning Applications via Soft Assertions

FSE 2025 paper introducing Soft Assertions to detect and trigger numerical instability bugs in ML applications.

  • Python
  • PyTorch
  • TensorFlow
GitHub
LLNL / Varity Across GPUs. HIP code generation for numerical testing.

GPU Numerical Testing – HIP Code Generation in Varity

Contributed to the LLNL Varity project by implementing HIP backend generation for GPU kernel testing, enabling cross-platform numerical consistency evaluation.

  • C++
  • HIP
  • CUDA
  • LLVM
GitHub
Language + geography Words in context. Evaluating hate speech detection with LLMs.

Hate Speech Detection with Geographical Context

Extended a class project to evaluate hate speech detection models using geographical metadata. Later adapted for large language model testing and published in 2025.

  • Python
  • Transformers
  • HuggingFace
  • Scikit-learn
GitHub
Computer vision Face + liveness Recognition and spoof detection for banking.

Face Recognition with Liveliness Detection

Developed a real-time face recognition and spoof detection system for FinTech applications, enhancing security for remote banking verification.

  • Python
  • OpenCV
  • Dlib
  • FaceNet
GitHub
Other repositories
  • Parliament Resource Planning – SSO Integration — Implemented Single Sign-On authentication for PRP system, enabling secure unified access across parliamentary resource modules.
  • MoneyGram SOAP API Integration — TODO: verify scope/details — inferred only from the repo name. Integrated a SOAP-based API with MoneyGram remittance services as part of banking platform work.
  • COM S 572 – Machine Learning Project — Implemented and evaluated ML algorithms for natural language processing tasks. Later extended for LLM experiments and paper publication.
  • Othello AI Agent — Developed an intelligent Othello game-playing agent using adversarial search algorithms and heuristic evaluation functions.
  • Mancala Game Simulation — Created a Mancala game engine with an AI opponent using minimax and heuristic strategies.
  • Ray Tracing Renderer — Built a ray tracing engine from scratch to render 3D scenes with reflection, refraction, and shading.
  • Nachos – Instructional Operating System — Extended the Nachos instructional OS to implement thread scheduling, virtual memory, and file system operations.
  • Channel Equalization for Wireless Communication — Implemented and tested channel equalization algorithms to improve signal quality in noisy communication environments.

## Teaching

Designing courses. Teaching through practice.

Teaching assistant in the Department of Computer Science at Iowa State University, contributing to course design, introductory programming, and artificial intelligence.

## Writing

Research notes & engineering write-ups

May 1, 2026

What I Want This Blog to Become

Academic papers are necessarily compressed. They describe the method, the evaluation, and the result, but they often leave out the messy middle: why a test design chan...

All posts →

## Contact

Let’s connect.

The fastest way to reach me is email.