MSIMohammad Sabik Irbaz

CLINICAL NLP · LLM SYSTEMS · APPLIED ML

AI that understands
people.

I’m Mohammad Sabik Irbaz. Call me Sabik.

I build language systems that connect how patients describe their health with how clinical systems understand it.

Explore my work ↘Download CV ↓

PhD researcher, George Mason University
Previously Leadbook · Omdena · Pioneer Alpha

Professional portrait of Mohammad Sabik Irbaz
RESEARCHER + ENGINEER

From the research question
to the working system.

01 / SELECTED WORK

Hard problems.
Human consequences.

My work spans patient language, rigorous evaluation, and the engineering needed to put models to use.

CLINICAL NLPONGOING RESEARCH

Turning patient language
into clinical meaning.

People rarely describe their health in the language of a medical terminology. I develop retrieval-augmented pipelines to connect patient narratives to RxNorm, SNOMED CT, and CPT-4 concepts.

My contribution
Hybrid retrieval, cross-encoder reranking, LLM-based selection, and text style transfer on multi-GPU HPC infrastructure.
Why it matters
More reliable language understanding is a foundation for useful patient-facing conversational systems.
RAGLangGraphClinical terminology
Research details in my CV ↗

RESEARCH APPROACH

01 / LANGUAGE

Patient-authored narrative

02
RetrieveFind candidate concepts
03
RerankResolve semantic similarity
04
Select & clarifyGround the interpretation
Standardized clinical concepts

Conceptual workflow · Research in progress

HEALTH TEXT SIMPLIFICATIONJBI · 2024

Making cancer education
easier to understand.

Co-first author of research introducing SimpleDC, a parallel corpus for digestive cancer education, and a learned reward for health text simplification.

Approach
LLM fine-tuning, reinforcement learning, and human-feedback-based rewards, evaluated on SimpleDC and Med-EASi.
Outcome
A published study and publicly available dataset supporting health text simplification research.
RLHFLLMsDataset creation
Read paper ↗Explore dataset ↗
CONVERSATIONAL AI EVALUATIONPREPRINT · 2026

Testing AI across
different patient voices.

Co-author of a patient simulation framework evaluating a conversational antidepressant decision aid across medical, linguistic, and behavioral profiles.

500simulated conversations in the study

The framework exposes performance differences across health literacy levels, helping characterize where conversational systems need improvement.

Patient simulationEvaluationHealth literacy
Read the preprint ↗

INDUSTRY / LEADBOOK

Building NLP for
business-scale data.

At Leadbook, I worked on company industry classification and job-title clustering, connecting model development with annotation operations and data quality.

1Mcompany records for language-model fine-tuning
30Kannotated examples for classification
148industry categories in the task

More things I’ve built

GitHub profile ↗
01 / NLP

A Transformer, from scratch ↗

PyTorch implementation for Bengali machine translation.

02 / CLASSIFICATION

141 genres. One description. ↗

Multi-label book classification with a Flask deployment implementation.

03 / END-TO-END ML

From model to interface ↗

20-class image recognition with Gradio and GitHub Pages integration.

Explore the earlier project archive ↗

2021–2026

News & milestones

A five-year record of research, publications, teaching, and professional growth.

SERVICE

Reviewed research for Communications Medicine, a Nature Portfolio journal

RESEARCH

Started PhD research at George Mason University and joined the Lybarger Language Lab

TEACHING

Began teaching applied data science as an instructor at MasterCourse

INDUSTRY

Joined Leadbook as a Data Scientist to build production NLP systems at business-data scale

EDUCATION

Completed a BSc in Computer Science and Engineering at the Islamic University of Technology

02 / EXPERIENCE

Research rigor.
Delivery mindset.

Experience across academic research, production ML, and international collaboration.

Full experience in my CV ↗

2023 – PRESENT

George Mason University

Graduate Research Assistant · Lybarger Language Lab

Clinical concept normalization, health text simplification, conversational AI, and information extraction. Advised by Dr. Kevin Lybarger.

2021 – 2023

Leadbook

Data Scientist

Developed and maintained NLP pipelines for industry classification, job-title clustering, and lead-generation data quality.

APR – JUN 2021

Omdena

Lead Machine Learning Engineer

Led recommendation and web deployment tasks on the Equilo gender-equality project, collaborating with 50 engineers across 20 countries. Stack included GCP, Django, React, and Firestore.

JAN – OCT 2021

Pioneer Alpha

Lead Machine Learning Engineer

Supervised and delivered client projects in multilingual conversational agents, Bangla license-plate recognition, and computer vision.

2019 – 2020

Nascenia

Software Engineer Intern

Contributed to a government VAT automation project using Ruby on Rails.

03 / PUBLICATIONS

Ideas, put on record.

Google Scholar ↗
Forthcoming & ongoing work

Natural Language Processing of Clinical Data. Co-author, Springer Handbook of Data Engineering. Accepted; publication expected October 2026, per CV.

Medical concept normalization in patient narratives. Style transfer, retrieval-augmented generation, and conversational clarification. Work in progress; PCORI-funded research.

Handwritten report information extraction. OCR and vision-language-model research; poster presented at an AMIA 2024 Symposium workshop.

Additional manuscripts and the complete publication list are in the CV.

04 / ABOUT

Curiosity,
with a purpose.

I’m interested in the gap between a model that performs well in an experiment and a system that works for the people using it.

I’m a PhD student in Information Technology at George Mason University, concentrating in Machine Learning Engineering. My research with Dr. Kevin Lybarger focuses on patient-centered language technologies.

Before GMU, I studied Computer Science and Engineering at the Islamic University of Technology in Bangladesh and worked in applied machine learning. I also collaborated with Fordham’s Human Centered AI Research Lab on public-policy NLP.

Meet Lybarger Language Lab ↗

What I work with

Language & learning

Python · PyTorch · Hugging Face · fastai · TensorFlow · RAG · RLHF

Systems & deployment

LangChain · LangGraph · Flask · Git · GCP · AWS · Azure · Multi-GPU HPC

Clinical language

SNOMED CT · RxNorm · CPT-4 · Concept normalization · Text simplification

EDUCATION

PhD · Information Technology
George Mason University
2023 – 2028 expected

BSc · Computer Science & Engineering
Islamic University of Technology
2017 – 2021

TEACHING

Making ML approachable.

Data Science Instructor at MasterCourse (2022–2026) and Head Course Instructor at Amar iSchool (2021), developing lectures, assignments, and projects for aspiring practitioners.

RECOGNITION

Two challenge wins.

Champion of the Nurse Care Activity Recognition Challenges in 2020 and 2022. OIC scholarship recipient; finalist in the AI for Bangla Challenge 2021.

SERVICE

Contributing to the field.

Reviewer for Communications Medicine (2025), WACV (2023), and EMNLP (2022).

Earlier talks & presentations ↗

05 / LET’S CONNECT

Building language AI
for real-world problems?

I’d welcome a conversation about clinical NLP, applied research, and ML engineering opportunities.

mirbaz@gmu.edu ↗
LinkedIn ↗GitHub ↗Google Scholar ↗Download CV ↓