Ryan Azim Roup
Available for AI & Full-Stack Roles
BRAC University CSE • 3.75 CGPA
Mohammadpur, Dhaka • Remote Available

Architecting Intelligent Systems & Modern Software Products.

Hi, I’m Ryan Azim Roup. I’m an AI/ML & Full-Stack Developer building AI-powered applications, modern web platforms, and automation solutions.

Explore ProjectsDownload CV
3.75BRAC University CGPAB.Sc. in Computer Science and Engineering
Swin + U-NetMultimodal AI ThesisPrivacy-preserving federated deep learning framework
Full-StackProduction CapabilityEnd-to-end architectures from models to modern web apps
runtime://applied-intelligence.sys
ACTIVE
Undergraduate ThesisDSpace Paper

Uncertainty-Aware Federated Multimodal Dual-Task Learning

Privacy-preserving skin lesion diagnosis combining Swin Transformer and Attention U-Net with adaptive uncertainty-aware client aggregation.

Backbone:Swin Transformer (Hierarchical)
Segmentation:Attention U-Net Dual-Branch
Aggregation:Adaptive (Entropy & Macro-F1)
Privacy Constraint: Zero raw data exchange
Federated LearningPyTorchSwin TransformerAttention U-NetMedical AI
ryan@bracu:~$Target: Full Intelligent Product
01 // Software Engineering & Research•Portfolio Showcase

Projects

Featured software products and computer vision research spanning distributed deep learning, full-stack ecosystems, and interactive simulation.

AI Research
Undergraduate Thesis
ID: federated-skin-lesion

Uncertainty-Aware Federated Multimodal Dual-Task Learning for Privacy-Preserving Skin Lesion Diagnosis

Undergraduate Thesis • Federated Learning & Computer Vision

A privacy-preserving distributed deep learning framework combining Swin Transformer and Attention U-Net for joint classification and segmentation of dermatological images without sharing raw patient data.

Engineered an uncertainty-aware dual-branch model combining hierarchical Swin Transformer with an Attention U-Net for joint skin lesion classification and lesion boundary segmentation under strict privacy constraints.
Implemented uncertainty-aware client aggregation to adaptively weight client model updates based on Macro-F1 performance and predictive entropy uncertainty.
Guaranteed zero raw clinical data transmission while outperforming traditional FedAvg and standalone single-task baselines on heterogeneous multi-client medical datasets.
ArchitectureSwin + Attention U-Net
AggregationEntropy & Macro-F1
Privacy LayerZero Raw Data Shared
PythonPyTorchSwin TransformerAttention U-Net+2
Paper
Full-Stack
Production App
ID: ghaughau-pet-adoption

GhauGhau — Full-Stack Pet Adoption Platform

Full-Stack Web Ecosystem & Community

An end-to-end pet rescue and adoption web platform connecting prospective adopters with shelter caretakers through streamlined matching workflows, detailed pet profiles, and secure management.

Architected a responsive, modern web application utilizing Next.js App Router, Tailwind CSS, Node.js backend, and MongoDB document database.
Implemented comprehensive listing workflows, photo galleries, pet filter taxonomies, and adopter inquiry management.
Designed responsive and accessible interfaces optimized for mobile touch interactions and fast page load speeds.
StackNext.js + MongoDB
UI LayerTailwind CSS Mobile-First
DeploymentProduction Cloud
Next.jsReactNode.jsTailwind CSS+2
Live Site
Systems & 3D
Interactive Game
ID: roadrash-3d

Road Rash with Obstacles — RoadRash3D

3D Game Engine & Physics Simulation

A 3D vehicular combat simulation built in Python using OpenGL, featuring real-time collision detection, enemy pursuit algorithms, and shooting projectile mechanics.

Constructed a 3D car-obstacle environment using PyOpenGL with custom camera viewpoints and perspective projections.
Engineered autonomous adversary vehicles programmed with target-tracking logic to pursue the player's position.
Implemented bounding-box collision detection for obstacles and projectile firing mechanics to neutralize oncoming threats.
RenderingPyOpenGL 3D
MechanicsCombat & Pursuit AI
EnvironmentReal-time Simulation
PythonPyOpenGL3D GraphicsGame Physics+1
Code
Commercial Web
Client Frontend
ID: one-scoop

One-Scoop — Commercial Brand Experience

High-Performance Brand Frontend

Modern, responsive commercial website designed and developed for an artisanal ice cream business, highlighting product flavor storytelling and responsive brand interaction.

Designed and coded responsive, mobile-first frontend with semantic HTML5, modern CSS3 styling, and JavaScript interactivity.
Structured product catalog showcase and visual flavor menus designed to maximize visitor engagement.
Optimized asset loading and cross-browser rendering for smooth consumer browsing across all screen sizes.
PlatformCommercial Brand Web
PerformanceFast & Mobile First
InterfaceSemantic Responsive UI
HTML5CSS3JavaScriptResponsive Design+1
Live Site

Looking for intelligent software product engineering?

Available for full-time engineering roles, AI research collaborations, and technical co-founding.

Initiate Collaboration →
02 // Technical Competence•Skills & Architecture

Capabilities & Tech Stack

Cross-disciplinary software engineering spanning deep learning research, modern full-stack web platforms, and automated robotic data pipelines.

AI & Machine Learning

Core Specialty
8 Tools

Deep learning research, computer vision pipelines, and uncertainty-aware federated learning architectures.

PyTorchSwin TransformerAttention U-NetFederated LearningComputer VisionLLM ApplicationsPrompt EngineeringMachine Learning Foundations

Full-Stack Development

Production Stack
8 Tools

Modern web applications built with type-safe architectures, responsive interfaces, and scalable API layers.

Next.js (App Router)ReactTypeScriptNode.jsTailwind CSSMERN StackRESTful APIsHTML5 / CSS3 / JavaScript

Databases & Cloud Platforms

Data Layer
6 Tools

Persistent data storage, authentication systems, cloud hosting, and continuous deployment workflows.

MongoDBMySQLVercelClerk AuthenticationGit / GitHubPostman API Testing

Automation & Systems Engineering

Robotic Pipelines
8 Tools

Robotic process automation, spreadsheet ETL ingestion, low-level systems programming, and 3D graphics.

Python AutomationSelenium WebDriverWorkflow OptimizationPyOpenGL 3D GraphicsC ProgrammingVerilog HDLAssembly LanguageRobotics & Microprocessors
Core Engineering Philosophy
01 //

Research-to-Production

Translating cutting-edge machine learning research (federated vision, Swin transformers) into operational software products.

02 //

Full-Stack Velocity

Engineering responsive web applications from database schemas to polished design-system user experiences in Next.js.

03 //

Robotic Automation

Developing robust Selenium scripts and Python data pipelines that eliminate recurring manual workflows and human errors.

03 // Track Record•Career & Credentials

Experience & Education

Applying intelligent automation systems, full-stack web platforms, and academic machine learning research in production and competitive environments.

Work & Internship History

Industry Internship
Industry Verified

Tech Intern — Automation & Data Systems

Krishi Shwapno

Mar 2026 – Jun 2026Dhaka, Bangladesh

Spearheaded internal workflow and robotic process automation, engineering automated ingestion pipelines that connected raw operational spreadsheets with administrative dashboards.

Engineered automated Python and Selenium WebDriver scripts to parse, validate, and inject spreadsheet datasets into central dashboard portals.
Eliminated dozens of hours of recurring manual administrative data entry, cutting human data-entry error rates to near zero.
Optimized internal data automation workflows and established structured validation error-handling routines for administrative personnel.
PythonSelenium WebDriverData AutomationSpreadsheet ETLWorkflow OptimizationProcess Engineering
View Certificate
Client Engineering

Freelance Web Developer

Self-Employed

Nov 2024 – Feb 2025Remote

Built functional, high-performance, and responsive commercial web platforms for independent clients using modern full-stack web technologies.

Delivered production-ready client websites tailored to brand identities, consumer onboarding, and commercial workflows.
Constructed modular UI component systems using modern JavaScript, React, Next.js, and Tailwind CSS.
Optimized site responsiveness, web accessibility, and performance metrics across desktop and mobile devices.
Next.jsReactJavaScriptTailwind CSSNode.jsClient Communication
Mentorship & Teaching

Private Technical Tutor

Self-Employed

Jan 2023 – Dec 2023Dhaka, Bangladesh

Provided academic and programming instruction, breaking down complex algorithmic and computing concepts for students.

Guided students through core computer science principles, mathematical foundations, and structured programming logic.
Tailored individualized curricula and hands-on exercises to strengthen analytical and problem-solving mindsets.
Technical InstructionComputer Science ConceptsProblem SolvingCommunication

Education & Competitions

Jan 2022 – Feb 2026

BRAC University

Bachelor of Science in Computer Science and Engineering

CGPA: 3.75 / 4.00

Focus on Artificial Intelligence, Machine Learning, Computer Vision, and Software Engineering. Author of Undergraduate Thesis on Uncertainty-Aware Federated Multimodal Learning.

View Thesis Paper
2025

CUET CSE Fest 2025

Datathon — National Competition

Certificate of Participation

Competed in high-intensity machine learning and data analytics challenges, analyzing complex datasets under timed constraints.

View Certificate
2019 – 2021

Academia

Cambridge A-Levels & O-Levels

A-Levels: GPA 5.00 / 5.00 • O-Levels: GPA 4.50 / 5.00

Strong foundational background in advanced mathematics, physics, and computing science.

Verified Academic Standing

BRAC University CSE Graduate • 3.75 CGPA

Strong theoretical grounding in algorithms, operating systems, and distributed AI.

04 // Get In Touch•Initiate Collaboration

Let's Build Something Intelligent.

Whether you are looking to build multimodal AI pipelines, scale a modern full-stack web platform, or automate business workflows—let's discuss how I can contribute to your team.

Available for Roles
Dhaka • Remote Worldwide
Primary Direct Channel

ryan.azim.roup@g.bracu.ac.bd

I typically reply within 24 hours. Feel free to reach out for engineering roles, AI consulting, or technical advisory.

Send an Email
PDF • Up to Date

Curriculum Vitae

Download my complete resume with academic thesis details, technical internship, and full skill proficiencies.

Download CV (PDF)