AI & NLP Engineer
Elsebaiy Mohammed is an AI & NLP Engineer specializing in speech processing, transformers, and multimodal AI systems. He is experienced in building and deploying end-to-end ML pipelines for ASR, NLP, and generative AI applications using PyTorch and HuggingFace. He has also contributed to published research and managed IT operations.
Expertise in developing and deploying advanced AI and Natural Language Processing solutions, including speech processing, transformers, and multimodal AI systems.
Proficient in building and deploying end-to-end machine learning pipelines, ensuring production-ready solutions with optimized performance and scalable architecture.
Experience in contributing to published research, optimizing model performance, and evaluating complex AI systems for real-world applications.
Built an API for evaluating pronunciation and fluency from speech recordings of predefined text, implementing Wav2Vec2 + CTC alignment with Goodness of Pronunciation (GOP) scoring.
Impact: Achieved ~88% scoring accuracy with ~10s inference latency in real-world usage.
Designed and built an end-to-end video dubbing system integrating ASR, NMT, and TTS pipelines for English to Arabic speech translation, including audio extraction, transcription, translation, speech synthesis, and video synchronization.
Impact: Enhanced output quality using punctuation restoration and Arabic diacritization models.
Fine-tuned Wav2Vec2 + CTC and Whisper on MaCro dataset (1200+ hours, ~24K speakers) to tackle domain shift across 17 unseen audio domains.
Impact: Achieved WER: 0.89 (Rank ~720 on a highly competitive benchmark).
Fine-tuned transformer models (mT5, BART, AraBERT) for abstractive Arabic summarization on large-scale datasets (LANS: 8M+ articles, WikiLingua) with custom preprocessing pipelines.
Impact: Evaluated using ROUGE metric, achieving 19.2 (Top 18 / 100 teams).
Co-authored a research paper on machine learning algorithms for colorectal cancer prediction, published in Springer BMC Artificial Intelligence.
Built and deployed end-to-end AI systems (ASR, NLP, speech applications) as a Freelance AI Engineer, delivering production-ready solutions.
Graduated with a B.S. in Computer Science & Artificial Intelligence from Benha University, achieving a GPA of 3.27/4.0 (Very Good).
Developed an AI Video Dubbing System as a Graduation Project, earning an Excellent with Honors grade.
Managed mission-critical IT operations, network infrastructure, and technical support as a Reserve Officer in the Egyptian Armed Forces.
Developed a pronunciation assessment tool achieving ~88% scoring accuracy with ~15s inference latency.