Skip to content
View sheyda2021's full-sized avatar
πŸ‘©β€πŸ’»
Working from home
πŸ‘©β€πŸ’»
Working from home

Block or report sheyda2021

Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
sheyda2021/README.md

Hi there, I'm Sheyda πŸ‘‹

AI/ML enthusiast turning research curiosity into practical, working systems

Email LinkedIn Kaggle Telegram GitHub


πŸ‘©β€πŸ’» About Me

πŸŽ“ B.Sc. in Computer Science Β· M.Sc. Student in Computer Engineering

I focus on building and evaluating machine learning systems β€” not just training models, but understanding when and why an architecture is the right choice for the problem and data at hand.

My recent work spans speech & audio classification, comparative architecture studies (CNN vs. attention-based models), and applied deep learning experiments β€” with an emphasis on honest evaluation: cross-validation, confusion matrices, and significance testing over cherry-picked metrics.


🧠 Areas of Interest

Core

  • Machine Learning & Deep Learning
  • Reinforcement Learning
  • Computer Vision
  • Speech & Audio Processing
  • Natural Language Processing

Applied

  • Data Science
  • Cybersecurity Analytics
  • Explainable AI (XAI)
  • MLOps

πŸ› οΈ Tech Stack

Languages

Python C++ MATLAB

ML / DL Frameworks

TensorFlow PyTorch Scikit-learn LightGBM

Data & Vision

NumPy Pandas OpenCV Matplotlib


πŸ“š Currently Learning

  • πŸ€– Large Language Models (LLMs)
  • βš™οΈ MLOps β€” moving models from notebook to production
  • 🧬 Advanced Deep Learning architectures
  • οΏ½οΏ½οΏ½οΏ½ Explainable AI (XAI)

🎯 What I'm Working Toward

I want to build intelligent systems that are not just accurate, but understood β€” where every architectural decision is backed by a clear, defensible evaluation. That mindset shapes how I approach every project: rigorous comparisons, transparent reporting of what worked and what didn't, and a continuous effort to sharpen both my ML intuition and engineering practice.


⭐ Thanks for stopping by β€” feel free to explore my repositories below!

Pinned Loading

  1. Network-Traffic-Intrusion-Detection Network-Traffic-Intrusion-Detection Public

    Machine Learning pipeline for network traffic intrusion detection using LightGBM and domain-specific feature engineering.

    Python 1

  2. Hepatic-Fat-Accumulation-Severity-Prediction Hepatic-Fat-Accumulation-Severity-Prediction Public

    Machine learning pipeline for hepatic fat accumulation severity prediction using clinical biomarkers, feature engineering, ensemble learning, and Macro F1 optimization.

    Python 1

  3. Heart-Disease-Risk-Prediction Heart-Disease-Risk-Prediction Public

    Machine learning pipeline for heart disease risk prediction using clinical features, feature engineering, ensemble learning, robust scaling, cross-validation, and MCC-based threshold optimization.

    Python 1

  4. Audio-Classification-CNN-vs-Conformer Audio-Classification-CNN-vs-Conformer Public

    A deep learning pipeline for audio classification comparing CNN and Conformer architectures, featuring Mel-spectrogram processing, data augmentation, and SNR robustness testing.

    Jupyter Notebook 1

  5. steel-quality-prediction- steel-quality-prediction- Public

    A machine learning pipeline for predicting steel quality scores using ensemble techniques (LightGBM, XGBoost, CatBoost) and feature engineering.

    Jupyter Notebook 1