Welcome to my GitHub portfolio! I am Muhammad Hamza, a Machine Learning Engineer with extensive hands-on experience in Deep Learning (DL), Natural Language Processing (NLP), and Generative AI. This repository showcases various projects and implementations that demonstrate my expertise across these domains.
- Built a long-text summarization model using BART and LED transformers.
- Focused on summarizing medical reports, leveraging
ccdv/pubmed-summarizationdataset. - Techniques: Transfer Learning, Transformers, PyTorch
- Lip-syncing model that maps audio to video in real-time using Wav2Lip.
- Optimized inference script for faster processing by reducing I/O overhead.
- Techniques: Computer Vision, Deep Learning, Model Inference Optimization
- Real-time speech transcription using Deepgram API for automatic speech recognition (ASR).
- Implemented efficient streaming from microphone input and synchronized text output.
- Techniques: ASR, Streaming Audio Processing, NLP
- Created AI-generated artwork using GANs for image generation.
- Focused on training stable GAN architectures and tuning hyperparameters for high-quality outputs.
- Techniques: Generative Adversarial Networks, Image Processing, DL
