Skip to content
View NazmusSaad's full-sized avatar

Block or report NazmusSaad

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
NazmusSaad/README.md

Hi there! 👋


I'm Md Nazmus Saad, a Computer Engineering student at the University of Toronto (PEY Class of 2028), passionate about building intelligent systems at the intersection of AI, software, and embedded robotics. My work spans fine-tuning large language models, deploying deep learning on edge hardware, developing real-time control for autonomous robots, and researching novel hardware optimization techniques for AI applications.

🚀 Featured Projects

🖥️ Software & Systems

  • Restaurant Wayfinder GIS (C++)
    GUI-based map & delivery route planner with Dijkstra/A* algorithms.
    [GitHub]

  • Optimized Game of Life Simulator (C)
    High-performance implementation of Conway’s Game of Life in C, focused on parallelism, SIMD vectorization, and cache-aware optimization.
    [Github]

  • Library Management System (Java)
    OOP-based system with search, borrow, return, and admin functionality.
    [GitHub]

  • Circuit Network Simulator (C++)
    Simulates voltage, resistance, and node-based circuit behavior.
    [GitHub]


🧠 Machine Learning & AI

  • MedLLaMA: Medical Assistant LLM

    • Fine-tuned LLaMA 2 with QLoRA on 25k medical Q&A pairs (92% accuracy).
    • Integrated LangChain + ChromaDB + DuckDuckGo to enable real-time RAG.
    • [GitHub]
  • Super-Resolution for Medical Imaging

    • Latent diffusion model with VAE + U-Net Lite (PyTorch) for 4× upscaling of chest X-rays.
    • Deployed with TensorRT + PyCUDA for 3.8× faster inference on edge GPUs.
    • [GitHub]
  • Pneumonia Detection CNN + Web App

    • Custom PyTorch CNN for binary chest X-ray classification (92% accuracy).
    • Deployed using Streamlit, OpenCV, and Pillow.
    • [Web App] | [CNN Code]

🤖 Embedded & Robotics

  • Air Hockey Game for RISC-V
    Built in C for a custom RISC-V architecture with graphics and input.
    [GitHub]

  • Autonomous Soccer Robots (UTRA)
    Built real-time control systems in Python & ROS2. Integrated YOLOv8 for visual strategy with PyTorch.


📊 Research & Algorithms

  • 2-Diophantine Approximation of Exp(x)
    Python tool with 15-digit precision for hardware-oriented optimization research.
    [GitHub]

📬 Connect with Me

LinkedIn GitHub

Pinned Loading

  1. Restaurant-Wayfinder-GIS Restaurant-Wayfinder-GIS Public

    An interactive, high-performance mapping tool built from scratch in C++. Restaurant Wayfinder helps users explore nearby restaurants and instantly find the fastest route to them using intelligent p…

    2

  2. utmist-vid-upscale/OBS-TensorRT-Filter utmist-vid-upscale/OBS-TensorRT-Filter Public

    OBS plugin filter that runs optimized ML inference every frame using TensorRT.

    C++

  3. Medical-Assistant-LLM Medical-Assistant-LLM Public

    🩺 A fine-tuned LLaMA-2 model for clinical instruction-following tasks. Trained on 200k medical instructions and deployed as an interactive Streamlit app. Evaluated using GPT-4-as-a-judge.

    Jupyter Notebook

  4. Game-of-Life-Simulator Game-of-Life-Simulator Public

    Engineered a high-performance version of Conway’s Game of Life, achieving 240× speedup over the baseline sequential implementation.

    C

  5. petrichor-2/image-upscaler petrichor-2/image-upscaler Public

    Jupyter Notebook 1

  6. utra-robosoccer/soccerbot utra-robosoccer/soccerbot Public

    Soccer playing robot representing Canada from University of Toronto

    C 154 9