FetalCareXAI
An explainable fetal-health classification application using a LightGBM model and LIME to turn predictions into interpretable evidence through an accessible Flask-based interface.
ENGINEER · RESEARCHER · BUILDER
I make complex systems easier to understand.
I’m Sabit Al Alfi, a computer science graduate building explainable AI systems and practical software—from research prototypes to products people can use.
BASEDhaka, Bangladesh
FOCUSExplainable AI
MODEResearch + Shipping
PROFILESabit Al Alfi
01 / ABOUT
FIELD NOTE 01
Clarity is a technical feature.
Research depth. Engineering range.
My work sits between understanding and building. I study why models make decisions, then carry that same first-principles mindset into databases, interfaces, graphics, and systems. The goal is not technology for its own sake—it is software that earns trust by being clear, useful, and well made.
02 / SELECTED WORK
Proof through projects.
A deliberately mixed selection: research-grade machine learning alongside shipped web and systems work.
Rhythm Wellbeing
A privacy-first wellbeing journal for logging time, understanding personal rhythms, receiving grounded AI reflections, and sharing opt-in insights through trusted Circles.
REMP
A Swin Transformer-based rare medicinal plant classifier achieving 98.46% accuracy. The work was developed into a published research paper.
Sabit Al Alfi Portfolio & CMS
A research-led portfolio paired with a secure owner CMS for managing projects, publications, posts, profile content, media, and site settings without compromising the code-owned design system.
03 / CAPABILITIES
Across the whole stack.
Languages & engineering
Python · C · C++ · TypeScript · JavaScript · SQL
Machine learning & analysis
PyTorch · TensorFlow · scikit-learn · Pandas · NumPy · Grad-CAM · LIME
Product & infrastructure
React · Astro · Node.js · Vite · Supabase · PostgreSQL · Git · Vercel
Data & research tools
Data cleaning · EDA · Matplotlib · Seaborn · LaTeX · Excel
04 / RESEARCH
Four published works.
A complete publication record across computer vision, explainability, agriculture, and environmental intelligence.
First author
IEEE · 2025REMP: A Swin Transformer-Powered Approach to Classifying Rare and Endangered Medicinal Plants
A customized Swin Transformer classified 16 rare and endangered medicinal plant species with 98.46% accuracy, supporting technology-assisted identification and conservation.
View publication ↗Second author
IEEE · 2025Explainable Deep Learning Paradigms for Nitrogen Deficiency Detection: Multi-Crop Assessment of CNN-Based Models for Agricultural Intelligence
An explainable multi-crop assessment using SE-ResNeXt and Grad-CAM, achieving 95.23% classification accuracy across four crop datasets.
View publication ↗Co-author
Springer · 2026Detection of Lemon Leaf Diseases Using Inception V3-Based Machine Learning Model
A comparative deep-learning study of lemon leaf disease classification in which Inception V3 achieved 86.94% validation accuracy, with explainability and augmentation identified as paths toward stronger precision-agriculture systems.
View publication ↗Co-author
Springer · 2026Forecasting Air Quality: A Comprehensive Survey of Air Pollution Prediction Methods and Applications
A survey of statistical, machine-learning, and deep-learning approaches to air-quality prediction, covering their comparative strengths, limitations, interpretability, and opportunities for real-time monitoring.
View publication ↗05 / CONTACT
Let’s build something that matters.
I’m currently looking for roles and collaborations in data science, machine learning, and AI engineering.
Open contact page →