Mahir Tajuar Akash

Machine Learning Engineer · Python · Django · NLP · MLOps

Mahir Tajuar Akash

Machine Learning Engineer with hands-on experience in deep learning, NLP, and computer vision. Passionate about building intelligent systems — from fine-tuning transformers for Bengali hate speech detection to deploying scalable ML applications on the cloud. Currently mentoring aspiring engineers while collaborating on business-driven AI solutions.

Get in touch Resume LinkedIn

Where I've Worked

Experience

  1. AI/ML Mentor and Project Co-ordinator

    Phitron, Programming Hero Limited

    Nov 2025 – Present Dhaka, BD
    • Teaching and conducting hands-on project sessions on Machine Learning, Deep Learning, and NLP.
    • Leading sessions on Automation, MLOps, and AI Engineering topics for aspiring engineers.
    • Co-ordinating learner projects and guiding them through end-to-end ML workflows.
    • Python
    • Machine Learning
    • Deep Learning
    • NLP
    • MLOps
    • Automation

  2. Intern — Executive, Business Department

    Qtec Solution Limited

    Dhaka, BD
    • Collaborating with technical teams to align business requirements with ML solutions.
    • Supporting the delivery of AI-driven products by bridging client needs and engineering execution.
    • Business Analysis
    • Stakeholder Communication
    • ML Solutions

Selected Work

Projects

Bengali Hate Speech Detection

BanglaBERT & XLM-R · Aug 2025 – Oct 2025

Fine-tuned BanglaBERT and XLM-R on Bengali text data collected via web scraping and Tesseract OCR, achieving ~90% classification accuracy. Built a FastAPI backend as a REST service and a Streamlit frontend for interactive access, containerized with Docker and deployed on AWS EC2.

  • BanglaBERT
  • XLM-R
  • Tesseract OCR
  • Hugging Face
  • FastAPI
  • Streamlit
  • Docker
  • AWS EC2

Health Insurance Fraud Prediction

NLP + ML for fraud detection

Implemented a multi-model approach combining NLP with classical ML — SVM, Random Forest, and Logistic Regression — alongside BERT embeddings to detect fraudulent health insurance claims with high accuracy.

  • Python
  • BERT
  • SVM
  • Random Forest
  • Logistic Regression
  • NLP

Multiple Disease Classification

CNN · VGG16 · MobileNet · ML models

Built deep learning models (CNN, VGG16, MobileNet) for image-based disease classification and benchmarked them against classical ML models (SVM, Random Forest, Logistic Regression) to identify the most effective classification strategy.

  • TensorFlow
  • Keras
  • CNN
  • VGG16
  • MobileNet
  • SVM
  • Random Forest

Research

Publications & Research

Ongoing Publication

Effective Health Care Insurance Fraud Detection and Interpretation: Integrating Machine Learning, Deep Learning, Language Models, and Explainable AI

M. T. Akash, S. Alam, Y. Hasan. A comprehensive study integrating classical ML, deep learning, language models, and explainable AI techniques for healthcare insurance fraud detection and interpretability.

Toolkit

Technical Arsenal

C C++ Python SQL Scikit-learn TensorFlow Keras PyTorch Pandas NumPy NLP Django Streamlit Selenium FastAPI Docker

Background

Education

BSc in Computer Science & Engineering

CCN University of Science and Technology

2022 – 2025

Recognition

Achievements & Activities

Languages

Spoken Languages

Bengali — Native English — Professional Working Proficiency

Say hello

Let's work together

Whether it's an ML engineering role, an NLP research collaboration, or a Django project — I'm always up for a conversation.

</