Open to collaboration

Hi, I'm

Yeasir Arafat

I build enterprise financial software at IFIC Bank PLC and publish research on privacy-preserving AI. My work connects customer onboarding, identity verification, core banking, and reliable data workflows, alongside research in privacy-preserving healthcare AI.

Banking-grade engineeringPublished AI researchSecurity-first systems

View My Work Read My Research Resume

Currently

Application Architect & Developer @ IFIC Bank PLC

Next.js.NETPyTorchBlockchain

5Peer-reviewed publications

27Citations (Google Scholar)

3.94Undergrad CGPA

189Bank branches running eKYC

189

bank branches on eKYC

plus 1,224 sub-branches, live since September 2026

citations

Google Scholar, h-index 4

CS courses assisted

programming, OOP, computer systems, and algorithms labs

peer-reviewed publications

2 journal articles (Elsevier, Wiley) and 3 conference papers

What I do

A practical mix of product, AI, and secure systems

I like projects where interfaces, data, models, and trust boundaries all have to work together.

Next.js, .NET, APIs

Product engineering

I design clear interfaces, build reliable backends, and keep the user workflow close to the code decisions.

PyTorch, CV, NLP

Applied AI systems

I prototype models, evaluate tradeoffs, and connect machine learning work to software that people can actually use.

Auth, encryption, blockchain

Security and trust

I care about data integrity, privacy, and operational constraints, especially in finance and healthcare contexts.

Mentorship, contests, teams

Teaching and leadership

I have led student engineering work, supported university courses, and organized community learning experiences.

Research

Peer-reviewed work in trustworthy AI

Five publications spanning privacy-preserving healthcare AI, secure decentralized systems, computer vision, and Bengali NLP.

All publications

Workflow of the decentralized medical image sharing framework: upload with feature hashing and encryption, retrieval with smart-contract verification

Journal ArticleFirst author2025

Decentralized Medical Image Sharing: A Blockchain Based Approach with Subject Sensitive Hashing for Enhanced Privacy and Integrity

IET Blockchain (Wiley), Vol. 5, e70009

DOI: 10.1049/blc2.70009

Lung CT scan beside its HiRes-CAM class activation map highlighting the regions FVCM-Net used for a malignant prediction

Journal Article2026

FVCM-Net: Interpretable Privacy-Preserved Attention Driven Lung Cancer Detection from CT Scan Images with Explainable HiRes-CAM Attribution Map and Ensemble Learning

Biomedical Signal Processing and Control (Elsevier), Vol. 112, Part C

DOI: 10.1016/j.bspc.2025.108719

Selected work

Case studies with real technical weight

A few projects that show how I connect product workflows, research ideas, and implementation details.

View all projects

eKYC Customer Onboarding Platform project illustration

Banking

2026In production since Sep 2026 · 189 branches & 1,224 sub-branchesApplication architect and developer (backend & frontend)

eKYC Customer Onboarding Platform

Digital account opening with NID OCR and maker-checker approval

Digitized end-to-end customer onboarding for a national bank, in production since 27 September 2026 across all 189 branches and 1,224 sub-branches, from NID upload to account creation in the CBS.

Onboarding channels

Frontend + backend

Engineering scope

Retry-aware customer and joint-account creation with explicit handling of uncertain CBS outcomes.
PostgreSQL-to-Oracle migration with reconciliation, rollback generation, and legacy-file access.
Identity verification, maker-checker/BAMLCO review, PDF forms, and masked integration logs.

Next.js 16.NET 8Oracle 19cCBS IntegrationData MigrationIdentity VerificationLDAP

Read engineering case study

Decentralized Medical Image Sharing project illustration

AI + Security

2024Published in IET Blockchain (2025)Researcher and full-stack engineer

Decentralized Medical Image Sharing

Blockchain + deep learning for tamper-proof healthcare data

Published as a first-author journal article in IET Blockchain (2025) with 98% average integrity correctness across three radiology datasets.

Security layers

Next + Flask

Core stack

DNN-based feature hashing for compressed and original image consistency.
AES-256 encryption before decentralized storage.
On-chain feature map and verification record design.

Next.jsFlaskMongoDBMachine LearningBlockchainIPFSAES-256

Source Paper (IET Blockchain)

Process

How I move from idea to usable system

The workflow is intentionally simple: understand the problem, shape the system, build the core, then harden the handoff.

Frame the problem

Clarify users, constraints, risks, and what success should look like before choosing the stack.

Shape the system

Map the data flow, core screens, API boundaries, model behavior, and security assumptions.

Build the usable core

Implement the smallest complete workflow with polished interactions and clear failure states.

Harden and explain

Tighten performance, accessibility, documentation, and handoff notes so the work can keep moving.

Current focus

I am intentionally building depth across enterprise engineering, applied AI, and trustworthy systems.

Building branch-led and self-led onboarding, retry-aware core banking integrations, and Oracle migration tooling at IFIC Bank PLC.
Sharpening AI engineering patterns for model-backed products.
Exploring trustworthy AI for healthcare and low-resource language systems.

Summa Cum Laude

B.Sc. in CSE, United International University

CGPA 3.94 / 4.00

Runner-Up

UIU Intra-University AI Contest

Fake news detection using ML

8th Place

ML Olympiad - CO2 Emissions Prediction

TF User Group North Bengal

4th Place (Public)

CUET ETE Day 2023 - ML Competition

Competitive ML challenge

Next conversation

Have a role, research idea, or product workflow that needs careful engineering?

I am open to full-time roles, research collaborations, and focused side projects.

Contact me See experience