Profile Image

About me

Hello! My name is Md. Mahfuzur Rahman, but you can call me Mahim. šŸ˜Š I am a Machine Learning engineer and prospective PhD student that is enthusiastic about bringing innovation to the market through academic brilliance and practical experience.. While I aim to make impactful contributions in the field, I recognize that pursuing higher studies is crucial for deeper expertise. This combination of academic knowledge and hands-on experience will best position me to develop innovative solutions that make a positive difference. I approach this journey humbly, knowing there is always more to learn and discover.

Skills

Experience

Education

  • Languages
    Python, C/C++, Java
  • Scripting
    LATEX
  • Database
    MS SQL
  • MS Office
    MS Word, MS Powerpoint, MS Excel
  • Architectural Paradigm for ML
    TensorFlow, PyTorch, Langchain, RAG(Retrieval Augmented Generation), Knowledge Graph
  • Silicon Orchard LTD.
    Junior Software Engineer (Machine Learning Team)
    • Contributed to a corporate project by implementing a RAG system and overseeing API deployment utilising FastAPI. Implemented the system on the companyā€™s proprietary server.
    • Engaged in a research project utilising the YOLO object detection model, which would bring value in better disaster assessment.
    • Engaged in an ongoing initiative that focuses on improving current operations. In this work, I used the fuzzy c-means clustering method, evaluated the clustering, and used the existing autoencoder model for text processing.
    • Contributing to developing a physics-guided clustering approach for vehicle imagery, focusing on sub-type level grouping (e.g., clustering different vehicle types within images).
    • Involved in developing vision language models; specifically, I fine-tuned a VLM using domain-specific subject images and assessed the quality of the generated images, moving beyond the conventional CLIPScore for evaluation.
  • Ahsanullah University of Sceience and Technology
    Bachelor Of Science In Computer Science and Engineering
    CGPA:2.849/4.00
  • BCIC college
    Higher Secondary School Certificate Examination
    GPA: 4.25/5.00
  • Monipur High School
    Secondary School Certificate Examination
    GPA: 5.00/5.00

Projects

CONVERTING SEMANTIC SEGMENTATION TO OBJECT DETECTION

  • Converted Semantic segmentations to object detection(bounding box annotations)
  • Used python OpenCV library for the main convertion task
  • Used the images from the dataset RescueNet by Rahnemoonfar et.al. for my task
  • Converted the coordinates of the mask objects to YOLO coordinate format
  • The bounding boxes were drawn considering the objects minimum area so that the bounding boxes donā€™t overlap much

Open in GitHub

ML LEARNING ECG DATASET

  • A General ML Learning classification task using supervised unsupervised and ensemble approach
  • Performed EDA; Explored the dataset and implemented five supervised models (logistic regression, decision tree, linear SVC, SVM, kā€‘nearest neighbors): One unsupervised approach (kā€‘means clustering); Three ensemble methods (Random Forest, Gradient Boosting, XGBoost).

Open in GitHub

APPAREL VGG NET

  • Used VGG16 as a pretrained model to train the Apparel Dataset from kaggle. Necessary comments to understand the code are also provided as it is a learning task.

Open in GitHub

APPAREL OWN MODEL PYTORCH

  • Used Pytorch framework to build a custom CNN Model using the ā€Apparel Datasetā€ from kaggle.

Open in GitHub

APPAREL OWN MODEL KERAS

  • Used keras to build a basic CNN model
  • Used Apparel Dataset from kaggle

Open in GitHub

APPAREL RESNET

  • Used Pretrained Model ResNet version 2 on Apparel Dataset From Kaggle to train

Open in GitHub

APPAREL INCEPTION NET

  • Used pretrained model Inception Net to train the image dataset
  • Used Apparel Dataset From Kaggle

Open in GitHub

Publications

publication_UAV

Unmanned Areial Vehichle

Diverse Applications of UAV Datasets in Segmentation, Classification, Detection, and Tracking

publication_Uddessho

Uddessho

An Extensive Benchmark Dataset for Multimodal Author Intent Classification in Low-Resource Bangla Language

publication_SOK

SOK

Exploring Hallucinations and Security Risks in AI-Assisted Software Development with Insights for LLM Deployment

publication_Fuzzy

Physical Fuzzy Rule Based Unsupervised News Article Clustering

My CV

CV Cover

View Full CV
Download CV

Extracurricular Activities & Achievements

Events

AUST CSE Carnival 1.0

AUST CSE ODYSSEY Spring 2022

Naimul Memorial Cricet League Spring 2022

Certificates

1 / 2
Image 1
Image 2
2 / 3
Image 1
Image 2
3 / 4
Image 3
Image 4
4 / 4
Image 5
❮ ❯

Contact Me

mahim1066@gmail.com

+88 019 14 603813

Successfully Submitted!