portfolio

Hi 👋, I'm Md Abdullah Al Hasib

A passionate Machine Learning (AI) Engineer from Bangladesh

About me

I’m a highly motivated and self-driven Machine Learning Engineer with a strong foundation in developing and deploying scalable AI solutions. I specialize in building end-to-end ML pipelines, deep learning applications, and real-world AI-powered systems across domains like NLP, Computer Vision, and Generative AI. I am passionate about open source and delivering AI solutions that make a meaningful impact.

I have Technical Experience and am Interested In

• Machine Learning
• Deep Learning
• Computer Vision
• Natural Language Processing
• Generative AI
• MLOPs

SkillSets

Programming Language: Python

Data Analysis: Numpy, Pandas, Matplotlib, Seaborn, Altair

Machine Learning: Scikit-learn, XGBoost, Ensemble Models

Deep Learning: Tensorflow, Keras, Pytorch, Transformers

Computer Vision: OpenCV, Ultralytics, Pillow, Scikit-image, Detectron2, YOLO

Natural Language Processing: NLTK, Gensim, TextBlob, Hugging Face, StandfordNLP

Generative AI: LangChain, langgraph, RAG, OpenAI Api, Hugging face, Pinecone (Vector Database), ChromaDB, Qdrant

MLOPs: Docker & Docker Compose, MLFlow, DVC, Apache Airflow, CLOUD(AWS)

Others: Git & Github, Streamlit, Fast API, Flask

Projects

Machine Learning

• Forest Cover Type Prediction
  • End-to-end ML pipeline (Data ingestion → Training → Evaluation) (Demo)
  • Tools: Scikit-learn, XGBoost, Pandas, Numpy
• IPL Score Prediction
  • End to end pipeline (Data ingestion -> EDA -> Feature Engineering -> Training -> Evaluation -> Deployment (Demo)
  • Regression model to predict the final score in live T20 matches
  • Tools: Pandas, Scikit-learn, Streamlit

Deep Learning

• Image Classification
  • Image Collection -> Preprocess -> Model Training -> Evaluation -> Inference
  • Tools: Pytorch, TorchVision, Matplotlib
• Brain Tumor (MRI) Detection
  • Data Collection -> Model Training -> Inference Test ->Evaluation
  • Tools: Pytorch, Torchvision, Numpy, Pandas, Matplotlib, Kaggle, MLFlow

Computer Vision

• Attendance through Face Recognition
  • Extract face from camera -> Recognize face -> append attendance to the Excel
  • Tools: Pandas, OpenCV, face_recognition, Excel
• License Plate Detection
  • Data Collection -> Processing -> Training -> Inference (Github Repo)
  • Tools: Ultralytics, YOLOV8, OpenCV, Pandas, Numpy, os, matplotlib
• Other Computer Vision Demo
  • Image Annotation Roboflow
  • Car & License Plate recognitation with Tracking
  • Vehicle Counting with Object Tracking
  • Adult & Child Detection with Yolov8
  • Instance Segmentation

Natural Language Processing

• Sentiment Analysis
  • Three approaches for sentiment analysis - (Vader, Textblob, Transformer Model) (Demo)
  • Created an APP with Streamlit
  • Tools: Vader, Huggingface, Textblob, Streamlit, Pillow
• Text Classification
  • End-to-end pipeline for Text Classification task
  • Tools: Scikit-learn, Pandas, Matplotlib, NLTK, Wordcloud, Joblib

Generative AI

• Q&A Chatbot
  • QnA chatbot through Gemini API and create an APP with Streamlit (Demo)
• Intent Based AI Search Application
  • E-commerce Search System through AI with safety Guardrails (Combined)
  • Tools: Huggingface, Langchain, RAG, Vector Database, Finetuning LLM, FastAPI
• RAG Project
  • Chatbot Application with RAG (able to add knowledge base through UI)
  • Tools: OpenAI, LangChain, Docker, Pandas, PymuPDF, FastAPI, Flask, Qdrant
• OpenNotebook API
  • OpenNotebook is an open-source Notebook LM that is decorated by Streamlit UI.
  • Convert the OpenNotebook into API
  • Tools: Docker, FastAPI, OpenAI, langchain, Langgraph
• Generative AI (Education Purpose)
  • I write a blog on Medium and make tutorials on YouTube. This repo is made for the learners.
  • Developed RAG, Quiz Test, Resume Checker for HR Projects
• STT(Whisper) Finetuning
  • End-to-end pipeline (Private Repo)
  • Dataset Creation (Extract Arabic Transcription and Audio from YouTube), create a dataset and upload to Huggingface Hub.
  • Data Ingestion from Huggingface -> Preprocess -> Training Whisper Model -> Evaluation
  • Tools: Huggingface, Transformers, AWS Sagemaker, Pytube, Pandas, Soundfile
• Voice Bot With RAG
  • Perform RAG based on user Query and convert the text output into audio through ElevanLabs (Backend API)
  • Tools: RAG, Vector Store, FastAPI, Elevenlabs, Hamsa API, LangChain
• AudioSense Project
  • Transcript Audio from Customers & Agents -> Analyze transcription & Audio -> Deployment (Demo)
  • Tools: AWS, OpenAI, Gemini, Sentiment Analysis, SqlAlchemy, Postgres, FastAPI, OpenNotebook, RAG, Docker Compose, CronJob

Web Automation

• Scraping Web Page
  • Extract Screenshots of the Car Information through VIN Number from CarfaxOnline.
  • Send the PDF to the Telegram Bot
  • Tools: Telegram, Selenium, Pandas, FastAPI, AWS(Windows-ec2 machine)

API

• Zip File Extraction API
  • Download Zip -> Extract Zip files -> Store AWS S3 bucket -> Provide file links
  • FastAPI, AWS, Pydantic, Uvicorn

Open Source Contribution

• ENG_TEXT_CLEANER: A Python package for cleaning the text from unnecessary characters like stopwords, punctuation, emojis, spelling mistakes and other preprocessing techniques.
• NoCodeTextClassifier: A Python package for automatically perform the text classification task including text preprocessing, training the model, evaluation and inference. This will help you to perform a text classification task without knowing the depth of the NLP methodologies. Also It will help you to perform hyperparameter tuning to find the best model for your dataset.

Education

Bachelor of Science, Computer Science & Engineering

Islamic University, Bangladesh

Connect with me: