Al Rafi Aurnob — AI/ML Engineer & Researcher

About Me

📋 Professional Summary

Motivated AI/ML Engineer with hands-on experience in deep learning, Generative AI, and real-world AI system development. Proficient in Python, PyTorch, TensorFlow, and Scikit-Learn, with applied exposure to LLM-based applications, RAG pipelines, vector embeddings, and NLP. Published researcher with 10+ peer-reviewed articles across Q1 journals and international conferences. Eager to contribute to AI & Innovation teams and build impactful intelligent solutions.

🔍 Research Interests

Deep Learning & Computer Vision
Federated Learning & Privacy-Preserving AI
Generative AI (GANs, Diffusion Models, LLMs)
Explainable AI (XAI) — Grad-CAM, SHAP
Multimodal & Hybrid Architectures
Medical Image Analysis & Clinical AI

PyTorch TensorFlow LangChain RAG Vector Embeddings Federated Learning XAI

Professional Experience

Research Data Analyst / Research Assistant

Advanced Machine Intelligence Research Lab (AMIR Lab)

Sep 2024 – Present

📍 Dhaka, Bangladesh

Designed privacy-preserving federated learning frameworks with homomorphic encryption for Parkinson's diagnosis & kidney imaging (IEEE Access, ICONIP 2024).
Developed explainable deep learning pipelines integrating Grad-CAM & SHAP for clinical interpretability in diabetes diagnosis, blood cell classification & cancer detection.
Engineered multimodal & hybrid architectures (CNN-Transformer, FedFusionNet) for oral cancer recurrence & ovarian cancer classification — published in Information Fusion (IF: 15.5).
Investigated generative AI (diffusion models, GANs) for data augmentation & fire detection; applied transfer learning (EfficientNet, BiT) to dermatological image classification.
Contributed to 10+ peer-reviewed publications across Q1 journals & indexed conferences.
Built biomedical data pipelines (PyTorch, Pandas, NumPy) reducing preprocessing time by 45%.

Data Analyst / Data Scientist Intern

ACI Motors Ltd

Feb 2024 – Jul 2024

📍 Dhaka, Bangladesh

Cleaned & analyzed crop yield, harvester GPS & credit transaction datasets using Python & SQL, reducing reporting turnaround by 40%.
Built interactive Power BI & Looker Studio KPI dashboards for 3+ operational teams, cutting manual report generation by 50%.
Designed a credit movement tracker in Python to flag transaction anomalies, improving financial data accuracy by 35%.
Implemented collaborative filtering recommendation system boosting product recommendation relevance by 38%.
Conducted geospatial analysis on harvester movement data, delivering insights that improved route optimization by 20%.

Assistant Investigator

Artificial Intelligence Research & Innovation Lab (AIRIL)

Nov 2023 – Present

📍 Dhaka, Bangladesh

Conducting advanced research in deep learning, federated learning & generative adversarial networks (GANs).

Co-Founder

Learnify Research Lab (LRL)

Dec 2023 – Present

📍 Dhaka, Bangladesh

Leading research initiatives in deep learning, federated learning & GANs; managing collaborative projects & mentoring team members.

Research Publications

10+ peer-reviewed articles — Q1 Journals & Indexed International Conferences

Journal IF: 15.5

FedFusionNet: Advancing Oral Cancer Recurrence Prediction through Federated Fusion Modeling

Al Rafi Aurnob, Sharia Arfin Tanim, Tahmid Enam Shrestha, Durjoy Mistry, M. F. Mridha

Information Fusion — Published

Published Link

Journal IF: 4.9

Explainable Deep Learning for Diabetes Diagnosis with DeepNetX2

Sharia Arfin Tanim, Al Rafi Aurnob, Tahmid Enam Shrestha, MD Rokon Islam Emon, M.F. Mridha, Md Saef Ullah Miah

Biomedical Signal Processing and Control — Published, 2024

Published Link

Journal IF: 3.4

Secure Federated Learning for Parkinson's Disease: Non-IID Data Partitioning and Homomorphic Encryption Strategies

Sharia Arfin Tanim, Al Rafi Aurnob, M. F. Mridha, Mejdl Safran, Sultan Alfarhood, Dunren Che

IEEE Access — Published, 2024

Published Link

Journal

A Comprehensive Framework for Forest Fire Detection Using Generative AI and Multimodal Deep Learning

Al Rafi Aurnob, Sharia Arfin Tanim, Tapu Biswas, M. F. Mridha, Mejdl Safran, Sultan Alfarhood, Dunren Che

Fire Technology — Published

Published Link

Journal

Enhancing Monkeypox Diagnostics: Exploring the Potential of EfficientNet and Big Transfer

Sharia Arfin Tanim, Kazi Tanvir, Al Rafi Aurnob, Md. Hasibur Rahman, Tasmia Binte Munir Maisha, Kamruddin Nur

Journal of Image and Graphics — Published, 2024 (Acceptance: 19%)

Published Link

Journal

Identification of Eye Diseases with Small Fundus Image Datasets Using a New Unsupervised Deep Learning Approach

Al Rafi Aurnob, Sharia Arfin Tanim, Farook Sattar

Frontiers in Signal Processing — Conditionally Accepted (under minor review)

Under Review Link

Conference

Explainable Federated Stacking Models with Encrypted Gradients for Secure Kidney Medical Imaging Diagnosis

Sharia Arfin Tanim, Al Rafi Aurnob, MD Rokon Islam, Md Saef Ullah Miah, M. Mostafizur Rahman, Mufti Mahmud

31st ICONIP — Published, 2024

Published Link

Conference

FusedMAC Framework Analysis for Enhanced Animal Condition Monitoring

Al Rafi Aurnob, Sharia Arfin Tanim, Tonmoy Mohajan, Kamruddin Nur

13th ICECE — Published

Published Link

Conference

Enhancing Blood Cell Classification by Applying Big Transfer and XAI

Sharia Arfin Tanim, Al Rafi Aurnob, Tahmid Enam Shrestha, Tamimul Alam, Kamruddin Nur

5th DOSCI — Published, 2024 (Acceptance: 20%)

Published Link

Conference

DeepResVit: A Hybrid Deep Learning Approach for Ovarian Cancer Classification with XAI

Al Rafi Aurnob, Tahmid Enam Shrestha, Md Sadi Al Huda, Shaif Ahamed Tamim, Ridwan Ahmed Arman, Md. Asraf Ali

ICICT 2024 — Published

Published Link

In Progress

Image-to-Image Decentralized Ovarian Cancer Classification Using Graph Attention Network

Ongoing Research

In Progress

Federated Learning-Driven Ovarian Cancer Classification Using Diffusion Models: A Novel Approach for Privacy-Preserving Medical Image Analysis

Ongoing Research

Key Projects

DeepResVit: Ovarian Cancer Classification

PyTorch · ResNet152 · ViT · Grad-CAM++

Hybrid deep learning combining ResNet152 & Vision Transformer on histopathology images — 98.65% test accuracy. Integrated Grad-CAM++ heatmaps for clinical interpretability, increasing clinician trust by 30%.

🏆 Published at ICICT 2024

AI Policy Chatbot (RAG / LLM)

Python · LangChain · ChromaDB · OpenAI API · NLP

Generative AI chatbot with RAG architecture over 50+ policy PDFs. ChromaDB vector embeddings achieved ~92% retrieval relevance, reducing document lookup time by 70%. Sub-second semantic search across 10,000+ sections.

AI-Powered Sales CRM

Python · Scikit-Learn · REST API · Automation

Automated lead scoring & persona generation pipeline improving qualified-lead identification by 38%. REST API endpoints for real-time lead ranking, reducing sales response time by 45%.

Credit Card Fraud Detection

Python · SMOTE · Scikit-Learn · SQL

Applied SMOTE on skewed transactional data achieving ~99.9% detection accuracy. Engineered transaction-level features increasing model recall by 22% on unseen fraud cases.

Market Basket Analysis & Customer Segmentation

Python · mlxtend · K-Means · Excel

Apriori algorithm for association rule mining — identified cross-sell pairs driving 25% increase in promotional revenue. K-Means clustering for targeted campaigns improving conversion rates by 18%.

Federated Learning for Parkinson's Classification

OpenFL · Flower · CKKS Homomorphic Encryption

Secure FL pipeline with CKKS encryption; improved Non-IID data robustness by 28%. Published in IEEE Access (IF: 3.4).

🏆 Published in IEEE Access

Core Skills & Technologies

💻 Programming & Scripting

PythonRSQL MySQLPostgreSQL

🧠 ML & Deep Learning

PyTorchTensorFlowKeras Scikit-LearnGenerative AILLMs RAG PipelinesVector Embeddings

📊 Data Analysis & Visualization

PandasNumPySQL Power BILooker StudioTableau MatplotlibSeabornAdvanced Excel

🔧 Tools & Platforms

GitDockerREST APIs JupyterGoogle ColabKaggle OpenAI APIHugging FaceLangChain ChromaDBJira

🔬 Research & Specialized

Federated LearningComputer Vision XAI (Grad-CAM, SHAP)Homomorphic Encryption GANsDiffusion Models Transfer LearningPrompt Engineering

🤝 Soft Skills

Data StorytellingStakeholder Reporting Cross-functional CollaborationResearch Writing Problem-SolvingAdaptability

Education

American International University – Bangladesh (AIUB)

B.Sc. in Computer Science & Engineering · Major: Information Systems

Jan 2020 – Apr 2025 · Dhaka, Bangladesh

Relevant Courses: Machine Learning, Computer Vision & Pattern Recognition, Data Mining, Probability & Statistics, Advanced DBMS, Data Structures & Algorithms

CGPA: 3.59 / 4.0

Cantonment Public School and College, Rangpur

Higher Secondary Certificate (HSC) — Science

2017 – 2019 · Rangpur, Bangladesh

GPA: 5.0 / 5.0

Birganj Pilot Government High School, Dinajpur

Secondary School Certificate (SSC) — Science

2017 · Dinajpur, Bangladesh

GPA: 5.0 / 5.0

Awards & Funding

🏅

Research grant from King Saud University, Riyadh, Saudi Arabia

Grant Number: RSPD2024R1027

Project: "Secure Federated Learning for Parkinson's Disease: Non-IID Data Partitioning and Homomorphic Encryption Strategies" — 2024