Ibrahim

Principal AI Engineer · Speech & ASR

Engineering conversational AI and speech recognition models

I specialise in Automatic Speech Recognition, self-supervised learning, and on-device AI deployment — with deep expertise in multilingual and multimodal ASR. PhD-trained, industry-proven at Recordesure, Speechmatics and Huawei, and based in Cambridge, UK.

Automatic Speech Recognition

End-to-end and hybrid ASR system design, multilingual model training, and evaluation across diverse acoustic conditions and languages including Arabic.

Self-Supervised Learning

Feature extraction with wav2vec 2.0, HuBERT, and AV-HuBERT; SSL-to-Kaldi pipeline integration; multimodal audiovisual representations.

On-Device AI Deployment

INT8 quantisation, model compression, and real-time inference for edge and mobile environments with strict latency and memory constraints.

PyTorch Kaldi lhotse icefall fairseq wav2vec 2.0 HuBERT AV-HuBERT Python C++ ONNX RAG Multilingual NLP Arabic ASR Microphone Arrays

Open to new opportunities in the UK and beyond.