Md. Mehenuf
Hossain Bhuiyan
I build ML systems, then make them explain themselves.
AI · Machine Learning · AI Automation & Operations
Data Analytics · Data Science
Computer Science & Engineering graduate from North South University, specialising in Artificial Intelligence & Data Science. I work end to end — data pipeline, model, evaluation, deployment — with a habit of shipping systems that are calibrated and auditable, not just accurate.
Research-minded, systems-fluent.
I'm Md. Mehenuf Hossain Bhuiyan, a Computer Science and Engineering graduate from North South University in Dhaka, specialising in Artificial Intelligence and Data Science.
My work sits where models meet production. I've trained a CNN from scratch and made its confidence scores trustworthy; distilled three heavyweight teacher networks into one edge-deployable student; and built a credit-risk pipeline whose every decision is committed on-chain so it can be audited after the fact. The through-line is that a model isn't finished when it's accurate — it's finished when someone else can verify it.
Alongside that, I run data work end to end: cleaning and wrangling messy tabular data, feature engineering, statistical analysis, and building the pipelines and dashboards that turn it into something a decision-maker can act on. I'm currently continuing research into few-shot learning for dermatoscopic image classification.
Outside the notebook I spent two years running esports operations for the NSU Computer and Engineering Club — sponsorship negotiation, 100+ volunteers, and the kind of SOP-writing that turns out to be the same instinct as building a reliable pipeline.
Roles I'm looking for
Open to full-time, internship, remote, and on-site.
- Artificial Intelligence
- Machine Learning Engineering
- AI Automation & Operations
- Data Analytics
- Data Science
- Computer Vision & Deep Learning Research
The formal record.
Most recent first.
B.Sc. in Computer Science and Engineering
Relevant coursework: Artificial Intelligence, Machine Learning, Pattern Recognition & Neural Networks, Image Processing, Statistics, Data Mining, Advanced Database Systems, Algorithm Design & Analysis, Theory of Computation.
Higher Secondary Certificate — Science
Secondary School Certificate — Science
Where the research happened.
Current work first — an ongoing investigation and the capstone that led to a published paper.
Undergraduate Researcher — Few-Shot Learning for Dermatoscopic Image Classification
- Investigating few-shot learning for skin-lesion classification under limited-label conditions.
- Designing pipelines that combine transfer learning and metric learning to handle class imbalance and domain shift.
- Evaluation via AUC, F1, and confusion matrices, with structured documentation of the experimental methodology.
Senior Design II Capstone — Decentralised Credit Scoring Protocol
Built the stacking-ensemble risk engine and the Node.js oracle bridge that became the basis of the ICIICE 2026 paper below. Full project write-up in Projects.
Tools I reach for, grouped by what they're for.
From training loops to smart contracts to spreadsheets — the stack behind the projects below.
Programming Languages
Machine Learning & Deep Learning
Explainable & Trustworthy AI
Data Analytics & Data Science
Frameworks & Libraries
Optuna
NumPy
pandas
OpenCV
ONNXtimmTorchVision
Flask
Gradio
Database & Backend
Blockchain & Decentralized Systems
Ethereum
ethers.js
IPFSAI Oracles
Infura
AlchemyMetaMaskPinata
Tools & Platforms
n8n
ClickUp API
Weights & Biases
Roboflow
Jupyter
Google Colab
Linux
Visualization & Reporting
Overleaf
Technical ReportingAcademic Writing
Six projects, six different constraints.
Each opens a full case study — the problem, the architecture, and the measured results, with the charts straight out of the repo.
CIFAR-10 Classification — SE-ResNet, CutMix & Calibration
95.48% top-1A custom SE-ResNet trained from scratch with no pretrained weights. Squeeze-and-Excitation attention, CutMix, and Focal Loss were each aimed at one measured failure mode — and cut cat/dog confusion by 17%.
Patient Care Operating System — ClickUp + n8n Automation
7 workflows liveA take-home Operations & AI Automation Specialist assessment: a ClickUp workspace engineered around a 60-use field budget, plus seven self-hosted n8n workflows covering everything the free tier can't do natively.
Decentralised Credit Scoring — Stacking Ensemble + Blockchain Oracle
AUC 0.984A risk engine for thin-file borrowers: SMOTE balancing, an XGBoost + MLP stacking ensemble, SHAP attributions, and a Node.js oracle that commits every score to Ethereum with the evidence pinned to IPFS.
XAI-Driven Lesion-Aware Multi-Teacher Knowledge Distillation
15.5 ms on CPUA U-Net-guided Inception teacher distilled into a MobileNetV3 student and quantized to INT8 — keeping teacher-grade F1 while running fast enough for edge deployment. Grad-CAM validates it looks at the lesion.
Relational DBMS — E-Commerce Perfume Store
3NF schemaA fully normalized MySQL schema with PL/SQL triggers and ACID-compliant transactions, behind a full-stack PHP storefront with cart management and a secure admin dashboard.
Flood Alert & Relief Coordination System
Geolocation routingA disaster-relief platform built on Firebase: live flood tracking, geolocation-based alert routing, and an admin dashboard for resource allocation and volunteer dispatch.
Peer-reviewed, presented, published.
Blockchain Meets AI-Powered Oracles and Decentralised Storage: A Secure Framework for Credit Scoring
A framework combining a stacking-ensemble risk engine, Ethereum smart contracts, and IPFS decentralized storage for transparent, tamper-evident credit scoring — examining how AI-powered oracles bridge off-chain inference and on-chain financial-inclusion systems.
XAI-Driven Lesion-Aware Multi-Teacher Knowledge Distillation for Robust Corn Leaf Disease Classification
With Satavisa Dey Borno. A lesion-aware dual-branch Inception teacher guided by U-Net mask proposals, distilled into a quantized MobileNetV3 student — combining explainability, compression, and selective prediction for field deployment.
Outside the notebook.
Running esports operations taught me the same lesson production ML does: structure the system before you scale it.
Deputy Director, E-Sports Wing
Led sponsorship strategy across three industry partners and commanded 100+ volunteers through structured SOPs and shift rosters.
Esports Organizer & Host
Ran the first flagship post-COVID LAN tournament at NSU, and hosted the university's first female-only Valorant tournament.
Organizer
Directed event moderation, logistics, and real-time technical troubleshooting across every tournament bracket.
Volunteer
Optimized registration pipelines and delivered technical support alongside senior coordinators.
Hiring for AI, ML, automation, or data? Let's talk.
I'm looking for roles across AI and machine learning engineering, AI automation and operations, data analytics, and data science. Email is the fastest way to reach me.