Skip to content
View bacemkarray's full-sized avatar

Block or report bacemkarray

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 is supported. This note will only be visible to you.
Report abuse

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

Report abuse
bacemkarray/README.md

Bacem Karray

Hey there, thanks for checking out my GitHub profile!

My name is Bacem, and I'm studying Computer Engineering at McMaster University.

Technical Interests:

  • Agentic Systems
  • Embodied AI (especially autonomous vehicles)
  • AI infrastructure
  • LLM/VLM inference
  • GPU programming

Hobbies:

  • Soccer
  • Swimming
  • Rock climbing
  • Flying
  • Travelling
  • Video games

Recent Projects

A C++ path tracer with both CPU and CUDA renderers. The project implements recursive ray scattering, antialiasing, gamma correction, depth-of-field camera controls, and multiple material types including diffuse, metal, and dielectric surfaces.

Tech: C++, CUDA, CURAND, Make, NVIDIA Nsight


A benchmark for parameter-efficient fine-tuning of a 7B language model on BBC XSum summarization. It compares LoRA and QLoRA across quality, training time, VRAM usage, and distributed training behavior.

Tech: Python, PyTorch, Hugging Face, LoRA, QLoRA, Accelerate


An autonomous hyperparameter tuning framework orchestrated with LangGraph. Parallel worker nodes tune individual neural-network hyperparameters while an LLM-powered coordinator analyzes results and decides whether to continue tuning or finalize training.

Tech: Python, LangGraph, PyTorch, Optuna, MLflow, OpenAI API


A minimal retrieval-augmented generation prototype built without high-level orchestration frameworks. It implements raw document parsing, chunking, embeddings, vector search, query rewriting, conversation memory, and response generation.

Tech: Python, ChromaDB, sentence-transformers, OpenAI API


A real-time American Sign Language to English translation tool built during York University's CTRL+HACK+DEL hackathon. It uses computer vision and machine learning to detect ASL gestures and convert them into English text.

Tech: Python, OpenCV, MediaPipe, TensorFlow, Flask


Tech Stack:

Python C C++ Bash Script PyTorch TensorFlow Keras scikit-learn NumPy CMake nVIDIA Google Cloud AWS Azure Redis Pandas Matplotlib mlflow Scipy Git GitHub

Pinned Loading

  1. agentic-hpt agentic-hpt Public

    Python

  2. asl-to-english-translator asl-to-english-translator Public

    CTRL+HACK+DEL 2024

    TypeScript

  3. face-tracker face-tracker Public

    Python

  4. transformer-playground transformer-playground Public

    Jinja

  5. exoskeleton-ai exoskeleton-ai Public

    Forked from McMaster-Exoskeleton/exoskeleton-ai

    Repository hosting code for Machine Learning Application used at McMaster Exoskeleton

    Python

  6. ray-tracing ray-tracing Public

    Cuda