My journey
My interest in technology grew from simple curiosity into a strong drive to build real, useful things, from web apps to intelligent systems that can understand data and language.
CSE @ North South University
Dhaka --:--
I love solving problems with technology. Right now I build web applications ML models reasoning pipelines clean databases
(01) About
I'm a computer science student who turns curiosity into working software, from full-stack web apps to models that read images and reason through language.
How I work
I break large problems into smaller, manageable pieces and approach them methodically. Whether it's designing a database, implementing an algorithm, or deploying a full-stack application, I care about clarity, structure, and maintainability. I learn by doing: building projects, experimenting with new tools, and turning ideas into working prototypes.
What drives me
At the core, I enjoy problem solving, especially how AI and machine learning can help in everyday digital experiences. I'm always looking for meaningful challenges, room to grow as an engineer, and teams that care about quality and impact.
(02) Skills
What I can do
01
Programming languages
Writing clean code for solving problems and building real-world applications.
Shipped
3+
projects across web, deep learning, and NLP, building since 2022.
02
Database & backend
Designing schemas, writing queries, and connecting applications to data storage.
03
Web apps & deployment
Building and deploying web applications, focusing on usability, performance, and reliability.
04
AI, ML & NLP
Exploring intelligent systems that learn from data and understand language to make useful predictions.
(03) Selected work
Things I've built
Mass Transport Ticketing System
A full-featured ticketing platform to improve mass transport in Bangladesh across metro, bus, and train. Online booking, seat selection, real-time updates, Google sign-in, user profiles, an admin dashboard, and reviews, built as a real upgrade over manual ticketing.
View on GitHubLung & Colon Cancer Classification
A deep learning system that classifies histopathological images into five cancer types through a web interface. Upload an image and get a prediction from the trained models.
View on GitHubReasoning While Decoding for LLMs
Analysis of decoding strategies for large language models, implementing a "Search Against Verifier" approach. A reasoning-while-decoding pipeline combines Chain-of-Thought prompting with Self-Consistency (majority voting) for more accurate outputs than plain inference.
View on GitHub(04) Contact