Skip to content
View M-Affan-Nazir's full-sized avatar
🤖
Probably developing an algorithm
🤖
Probably developing an algorithm
3 followers · 3 following

Achievements

Achievement: Pull Shark

Achievements

Achievement: Pull Shark

Block or report M-Affan-Nazir

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don't include any personal information such as legal names or email addresses. Markdown supported. This note will be visible to only you.
Report abuse

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

Report abuse
M-Affan-Nazir/README.md

Hi , I'm Affan

⚒️ I'm currently a third-year Computing Science and Artificial Intelligence Student at the University of Alberta ⚒️
🖥️ Passionate about Data Science, Machine Learning, and Algorithm Design 🖥️

github contribution grid snake animation

🚀 Languages and Tools:

python Node.js JavaScript dotnet C C++ C# mysql html5 git GitHub terminal bash zsh wing vim vscode pycharm photoshop msword msexcel mspowerpoint msonenote Figma Docker Azure Selinium dotnet TensorFlow arduino powerbi PyTorch

My Projects

📽️ Projects 🛠️ Tools 📚 Features
Facial Recognition CNN Python Keras TensorFlow
  • Built architecture using Python, Keras, TensorFlow to identify 138 facial landmarks.
  • Implemented 10 Convolution Layers with varying filter sizes and 3 Max Pools for prominent feature extraction.
  • Used Batch Normalization to improve training speed and LSTM for pattern recognition.
  • Achieved ±3 pixels accuracy using Adam optimizer with a learning rate of 0.0005 over 200 epochs.
Edmonton Traffic Statistics (ML & Statistical Inference) Python pandas NumPy scikit-learn Matplotlib Jupyter
  • End-to-end pipeline: EDA → cleaning → feature engineering → statistical inference → ML prototyping.
  • Ran Shapiro–Wilk, Levene, Kruskal–Wallis, and Wilcoxon tests with effect-size summaries.
  • Prototyped GLMs (incl. Negative Binomial) for yearly collision counts.
  • Reproducible notebooks and a final PDF report; raw data sourced via Edmonton Open Data links.
Transformer Encoder Python PyTorch
  • Built architecture using Python and PyTorch implementing Tokenization, Multi-Head Attention, and Positional Encoding.
  • Developed a classification model with 5 heads, 3 layers, and 8x8 matrices.
  • Achieved a 7% accuracy increase over traditional RNN models.
Global Debates Social Media Mobile App React Native JavaScript Node.js Express Socket.IO MongoDB
  • Developed the front-end with React Native and JavaScript.
  • Built the back-end using Node.js, Express, Socket.IO, and MongoDB.
  • Hosted the back-end on a VPS, enabling HTTP CRUD operations and real-time data transfer using sockets.
Polynomial Regressor and Logistic Classifier Python Scikit-Learn
  • Implemented Stochastic AdaGrad and mini-batch gradient descent to optimize Mean Squared Error (MSE) and Categorical Cross-Entropy (CCE) cost functions.
Human-Like Chatbot Python TensorFlow
  • Built a Seq2Seq model with 400 LSTM units and a vector size of 50 to generate text responses using an attention mechanism.
  • Managed encoder and decoder states effectively.
  • Achieved 93% accuracy on the Cornell Movie Dataset.
Advanced Stock Prediction Python TensorFlow Pandas
  • Developed a Recurrent Neural Network (RNN) for time series analysis with 4 layers and 200 LSTM units.
  • Achieved 95% accuracy on the test dataset.
Cancer Detection CNN Python TensorFlow
  • Developed a Convolutional Neural Network (CNN) with 2 convolutional layers and 2 max-pooling layers.
  • Achieved 98% accuracy on the test dataset.
Dynamic Currency Arbitrage Engine C++ libcurl JSON
  • Developed a C++ arbitrage system leveraging libcurl for API integration and nlohmann: json for data parsing.
  • Designed advanced graph structures with Dijkstra's Algorithm for optimal currency conversion pathways, achieving a 1% increase in real-world arbitrage gains through efficient route optimization.
  • Crafted an efficient algorithm leveraging HashMap structures that supported seamless integration of real-time market data, achieving response time improvements by reducing latency in currency conversion requests, improving lookup speed by 60%.
  • Created a custom Min Priority Queue using heap-based data structures to optimize shortest path calculations, ensuring high-efficiency and scalable computations, resulting in a 50% performance increase over arrays.
  • Integrated modular components for API fetching, data parsing, graph generation, and algorithm execution, demonstrating strong proficiency in C++, object-oriented design, and software architecture.
NAT Hackathon 2024 - BrainWave SVM: Real-Time Mental State Detection Python Scikit-Learn
  • Engineered a comprehensive solution for collecting and processing EEG signals through an innovative Python script, leading to the collection of 20 brainwave frequencies.
  • Normalized data and performed Two-tailed Welch’s T-Test to identify 4 statistically significant features.
  • Implemented and trained Support Vector Machine (RBF kernel, LOO cross-validation) on selected 4 features to identify individuals in excited states, achieving 90% accuracy of the trained model.

Pinned Loading

  1. facial-landmark-deep-cnn facial-landmark-deep-cnn Public

    Facial Landmark Detection using Deep CNN Model employs a hybrid architecture—combining CNNs, ANNs, and RNNs—to accurately detect 138 facial landmarks from images. Trained on the Helen Dataset, this…

    Python

  2. transformer-encoder transformer-encoder Public

    Transformer Encoder Model features a custom-built attention mechanism implemented from scratch in 'encoder_torch.py'. This flexible encoder leverages multi-head attention, layer normalization, and …

    Python

  3. raye-network raye-network Public

    RAYE is a global debates platform with a React Native front-end and a Node.js, Express.js, MongoDB, and Socket.io back-end. It features dynamic discussion groups, personalized tweet-like feeds, ann…

    JavaScript

  4. brain-wave-svm brain-wave-svm Public

    BrainWave SVM leverages real-time EEG data from a Muse S headband to classify states of excitement versus neutrality using advanced feature extraction and an SVM model. This repo includes data acqu…

    Jupyter Notebook

  5. edmonton-traffic-statistics edmonton-traffic-statistics Public

    Edmonton Traffic Statistics — Statistical inference & machine learning on Edmonton collisions, traffic volume, and weather, with reproducible notebooks, a clean data pipeline, and a concise report …

    Jupyter Notebook

  6. arbitrage-engine arbitrage-engine Public

    Dynamic Currency Arbitrage Engine is a high-performance C++ system that exploits real-time market data to identify optimal currency conversion routes using advanced graph algorithms and custom data…

    C++